[ { "id": 3870, "day": "2026-09-30", "type": "Talk", "start": "09:30", "end": "09:40", "track": [ "Apidays" ], "room": [ "Grand Hall 2", "Grand Hall 1" ], "title": "Opening Remarks", "speakers": [ { "name": "Mehdi Medjaoui", "title": "Author of Continuous API Management", "company": "Founder and Chairman of Apidays Conferences" } ], "facets": [], "agent_readiness_dimensions": [], "why": "Opening remarks, no abstract.", "abstract": "" }, { "id": 3992, "day": "2026-09-30", "type": "Workshop", "start": "09:35", "end": "10:00", "track": [ "GreenIO" ], "room": [ "Doddington Forum 2" ], "title": "Digital Impacts Schema and Taxonomy Workshop", "speakers": [ { "name": "Oliver Cronk", "title": "Founder", "company": "Architect Tomorrow / Cronk Advisory" } ], "facets": [ "contract_governance", "outside" ], "agent_readiness_dimensions": [], "why": "An open schema + taxonomy for non-carbon impacts is shared vocabulary; nothing in the rubric reads impact disclosure.", "abstract": "Organisations can measure the carbon footprints of their technology with increasing confidence, but carbon is only one dimension of technology's impact on the world. Water consumption, electronic waste, biodiversity loss, supply chain conditions, and the societal and individual harms caused by technology systems are all material impacts. Today there is no consistent, open, structured way for organisations to report on or share data about these impacts alongside their carbon data. Help be part of the definition of that standard in this workshop!" }, { "id": 4010, "day": "2026-09-30", "type": "Keynote", "start": "09:40", "end": "10:05", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "Trends Keynote: API Design Transforms DevOps Automation into Platform Products,", "speakers": [ { "name": "Abby Bangser", "title": "Founding Principal Engineer", "company": "Syntasso" } ], "facets": [ "developer_ergonomics", "contract_quality", "contract_governance" ], "agent_readiness_dimensions": [], "why": "Platform APIs as products: consumers, contracts, composition, change.", "abstract": "Platform teams have a clear mandate: curate the capabilities that matter across many teams, encode your organisation's context into them, and ship them as internal products. Automation and reliability engineering get you most of the way there. API design is the missing piece, and it decides whether what you've built feels like a product to the teams using it, or just another internal tool with a UI bolted on. The API community has spent decades figuring out how to design around consumers, contracts, composition and change. Platform engineering is only just starting to apply those lessons internally. This talk walks through how API thinking reshapes DevOps automation into real platform products, and makes the case that treating platform APIs as products, not afterthoughts, is the next stage of platform engineering maturity." }, { "id": 4031, "day": "2026-09-30", "type": "Keynote", "start": "09:40", "end": "10:05", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "Trends Keynote: The AI Gateway Playbook: prompt, responses & policy guardrails in production", "speakers": [ { "name": "Shruti Naik", "title": "Engineering Lead – Kubernetes", "company": "Lloyds Banking Group" }, { "name": "Fayzul Islam", "title": "Cloud Engineering Lead", "company": "Lloyds Banking Group" } ], "facets": [ "agent_readiness", "operational_transparency", "access_clarity", "regulatory" ], "agent_readiness_dimensions": [ "rate_limit_signal", "auth_clarity" ], "why": "Gateway quotas, auth, token limits and cost controls in a bank.", "abstract": "This talk explores how an API gateway can act as the control plane for Al workloads, enforcing prompt and response safety, policy-based access, and cost controls before requests ever reach an LLM. Using real production patterns, it shows how existing APl management capabilitiesrouting, quotas, authentication and observability, combine with Al-specific guardrails such as prompt/response sanitisation, semantic caching, and token-based limits to safely run agentic and GenAl use cases at scale." }, { "id": 3993, "day": "2026-09-30", "type": "Workshop", "start": "10:00", "end": "10:25", "track": [ "GreenIO" ], "room": [ "Doddington Forum 2" ], "title": "Scaling Sustainable IT & AI: Trust, Efficiency & ROI , via a Native Module ServiceNow", "speakers": [ { "name": "Matthieu Poulard", "title": "CEO", "company": "Aguaro" }, { "name": "Brian Minehane", "title": "Executive Advisor", "company": "AGUARO" } ], "facets": [ "access_clarity", "outside" ], "agent_readiness_dimensions": [], "why": "Cost + carbon steering; only the finops piece of Access Clarity touches it.", "abstract": "How do you scale AI without breaking Net Zero goals? Aguaro’s CEO and Brian Minehane (former Novartis Head of Sustainable IT) share a practitioner's blueprint for success. They will present a \"Measure-Action\" loop built on ServiceNow—leveraging the CMDB and Portal for trusted, effortless steering—while remaining open to other ecosystem integrations. From global scale to application level, learn to empower your teams from Build to Run, transforming sustainability into a lever for savings and resilience with an simple ServiceNow Upsell." }, { "id": 4011, "day": "2026-09-30", "type": "Keynote", "start": "10:05", "end": "10:30", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "From Open Banking to Autonomous Treasury: How APIs, ISO 20022 and Agentic AI are Creating Self-Driving Cash Management.", "speakers": [ { "name": "Dominic Inkoom", "title": "Director, Digital Channels, UK", "company": "Standard Chartered Bank, UK" } ], "facets": [ "regulatory", "contract_quality", "agent_readiness" ], "agent_readiness_dimensions": [ "agentic_commerce", "delegated_identity" ], "why": "ISO 20022 data conformance + agents initiating payments.", "abstract": "The convergence of APIs, ISO 20022 and Agentic AI is transforming treasury from a reactive function into an intelligent, autonomous capability. As financial institutions move beyond Open Banking, the next frontier is self-driving cash management, where systems can predict liquidity needs, initiate payments, optimise working capital, detect anomalies, and make real-time decisions with minimal human intervention. This session explores how rich ISO 20022 data, API-enabled connectivity and AI-powered agents are combining to reshape transaction banking and corporate treasury. Through practical industry examples and emerging market trends, attendees will gain insight into the technologies, architecture and strategic considerations that are enabling the next generation of intelligent financial services." }, { "id": 4026, "day": "2026-09-30", "type": "Keynote", "start": "10:05", "end": "10:30", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "Case Study Keynote: The Gateway Stack: Engineering LLMs and Agent Infra at Deliveroo and DoorDash", "speakers": [ { "name": "Saikrishna Desaraju", "title": "Senior Engineering Manager", "company": "Deliveroo (DoorDash)" }, { "name": "Jacob Lever", "title": "Senior Staff Software Engineer", "company": "Deliveroo" }, { "name": "Ruwan Geeganage", "title": "Staff Software Engineer", "company": "Deliveroo" } ], "facets": [ "operational_transparency", "agent_readiness" ], "agent_readiness_dimensions": [], "why": "Internal agent platform; reliability and cost. Weak public-surface signal.", "abstract": "Building an in-house Agentic AI platform is a crucial investment for our company to accelerate innovation and be at the forefront of the changing landscape in the delivery industry. It is equally important to build the foundational building blocks and frameworks to avoid repetitive and siloed work in the product teams with a strong focus on cost optimisation and reliability. We would like to share our success stories and provide the real-world impact of Agentic AI systems." }, { "id": 4012, "day": "2026-09-30", "type": "Panel discussion", "start": "10:50", "end": "11:15", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "Boomi Panel: State of the API Economy 2026State of the API Economy 2026", "speakers": [ { "name": "Mark Boyd", "title": "Director", "company": "Platformable" }, { "name": "Markus Müller", "title": "Global CTO - API Management", "company": "Boomi" } ], "facets": [ "access_clarity", "discoverability" ], "agent_readiness_dimensions": [], "why": "State of the API economy: the market-level read the catalog-wide score produces.", "abstract": "Join FOST Analyst Mark Boyd and Markus Müller, Global Field CTO from Boomi launch the API Economy in the Age of AI State of tyhe Market report 2026 Edition, and discuss key trends, forward-pushing predictions, and key insights on how API practitioners can be prepared to build their customer base and adapt their infratsructure in 2027." }, { "id": 4027, "day": "2026-09-30", "type": "Talk", "start": "10:50", "end": "11:15", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "The Blueprint for Agent-Ready APIs in Fintech", "speakers": [ { "name": "Yu Zhen Koh", "title": "Software Engineer", "company": "Wise" } ], "facets": [ "agent_readiness", "contract_quality", "developer_ergonomics", "operational_transparency" ], "agent_readiness_dimensions": [ "spec_presence", "mcp_server", "error_semantics" ], "why": "Machine-readable contracts, MCP, LLM docs, actionable errors, disciplined versioning. Close to the Agent Readiness rubric itself.", "abstract": "AI is enabling teams to do more with less, but agentic AI is only as useful as the APIs it can safely discover, understand and execute. This talk explores how machine readable contracts, MCP and LLM optimised documentation, actionable errors and disciplined versioning help developers build agent driven fintech workflows with the visibility and control production systems required." }, { "id": 4040, "day": "2026-09-30", "type": "Talk", "start": "10:50", "end": "11:15", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "The Holiday Checkout Has No Human", "speakers": [ { "name": "Shaked Vax", "title": "Head of API Security", "company": "Thales" } ], "facets": [ "agent_readiness", "operational_transparency" ], "agent_readiness_dimensions": [ "consent_identity", "agentic_commerce", "agent_identity_declared" ], "why": "Telling trusted shopping agents from bots.", "abstract": "Autonomous shopping agents will reshape holiday commerce and amplify fraud. This session reveals how to secure API journeys by separating trusted automation from bot abuse." }, { "id": 4047, "day": "2026-09-30", "type": "Workshop", "start": "10:50", "end": "11:40", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "Even an Intern with ChatGPT Can Do It: The New Reality of API Attacks", "speakers": [ { "name": "Paul Fearns", "title": "API Security Systems Engineer", "company": "Akamai" } ], "facets": [ "operational_transparency", "regulatory" ], "agent_readiness_dimensions": [ "auth_scheme_strength" ], "why": "AI-accelerated attacks; the published security + disclosure posture.", "abstract": "API attacks no longer require deep expertise. This live demo shows how simple techniques and AI can analyze, exploit, and accelerate API attacks—revealing why API security is now a board-level issue." }, { "id": 3994, "day": "2026-09-30", "type": "Workshop", "start": "11:00", "end": "11:25", "track": [ "GreenIO" ], "room": [ "Doddington Forum 2" ], "title": "What does AI emissions data look like in 2026?", "speakers": [ { "name": "Julia Lozano", "title": "GreenOps Specialist", "company": "Greenpixie" }, { "name": "Ross Hunter", "title": "Data Scientist", "company": "Greenpixie" } ], "facets": [ "access_clarity", "outside" ], "agent_readiness_dimensions": [], "why": "Token usage -> energy/carbon/water. Providers expose different usage data: a disclosure gap.", "abstract": "AI emissions are still a black box. Providers expose different usage data, infrastructure is hidden, and there's no single standard for turning AI activity into a meaningful emissions number. This session shows what AI emissions data can look like in 2026: how token usage translates into estimates of energy, carbon and water, and one consistent methodology applied across AI providers, SaaS tools and agents. We'll see how this data drives better AI decisions and cuts cost and emissions, then bring it to life with a live demo: the footprint of a Claude Code prompt, calculated on stage." }, { "id": 4028, "day": "2026-09-30", "type": "Talk", "start": "11:15", "end": "11:40", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "The Anatomy of an Agentic AI Sandbox", "speakers": [ { "name": "Ludovic Pourrat", "title": "API Architect & API Catalyst", "company": "Lombard Odier Group" } ], "facets": [ "developer_ergonomics", "agent_readiness", "contract_quality" ], "agent_readiness_dimensions": [ "mcp_server", "dry_run_mode", "well_known_catalog", "openapi_examples" ], "why": "Mocks, portal, MCP gateway + registry, synthetic data, workflow definitions.", "abstract": "As AI agents evolve from simple assistants into autonomous systems capable of discovering tools, executing workflows, and making decisions, traditional API developer portals and sandboxes are no longer enough. AI agents require an environment designed specifically for experimentation, evaluation, and safe integration. This talk dissects the anatomy of an Agentic AI Sandbox, a platform that combines the best of API developer experience with AI-native capabilities. Starting with mocked APIs and developer portals, we layer on an MCP gateway that exposes APIs as AI tools, an MCP registry for discovery, high-fidelity synthetic data for safe testing, and workflow definitions that enable agents to understand and execute real business processes. An Agentic AI Sandbox isn't just an API sandbox for AI, it's a developer acceleration platform that enables teams to build, test, and iterate AI agents at the speed of thought." }, { "id": 4088, "day": "2026-09-30", "type": "Talk", "start": "11:15", "end": "11:40", "track": [ "Apidays" ], "room": [ "Grand Hall 2", "Grand Hall 1" ], "title": "The story of API traffic management: from WAF to MCP", "speakers": [ { "name": "Irina Kamalova", "title": "VP Lead Software Engineer", "company": "JPMorgan Chase" } ], "facets": [ "operational_transparency", "agent_readiness" ], "agent_readiness_dimensions": [ "mcp_server", "rate_limit_signal" ], "why": "30 years of traffic management, WAF to MCP.", "abstract": "This talk traces the turbulent 30-year evolution of how we control the flow of data, moving from simple digital \"traffic lights\" to the intelligent \"brain\" of the modern enterprise. We begin in the late 90s, where the Dot-com Boom forced a transition from basic hardware to the software-defined world of Nginx and HAProxy. This era birthed the Reverse Proxy, transforming how we handled load and reliability. As complexity exploded, the industry moved toward dedicated management; we saw the rise of the first API prototypes, the community-driven power of Konga, and Google’s enterprise-grade solidification of the market. With the arrival of Kubernetes, the landscape fractured into a crucial duality: Service Mesh for internal \"East-West\" traffic and API Gateways for external \"North-West\" exposure. However, the dawn of Generative AI has rendered static routing obsolete. We conclude with the Model Context Protocol (MCP)—the final frontier where traffic management shifts from merely moving bytes to orchestrating \"intelligence\" and \"context\" for autonomous agents." }, { "id": 4041, "day": "2026-09-30", "type": "Workshop", "start": "11:15", "end": "12:05", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "Let’s Launch a Rocket: AI-Powered API & MCP Bootcamp", "speakers": [ { "name": "Anthony Virard", "title": "Senior Developer Advocate", "company": "Postman" } ], "facets": [ "developer_ergonomics", "contract_quality", "agent_readiness" ], "agent_readiness_dimensions": [ "mcp_server", "openapi_examples" ], "why": "Spec -> collection -> code -> MCP with Postman Agent Mode.", "abstract": "With the rise of Large Language Models, AI copilots, and autonomous agents, developers can now design, test, and iterate on APIs with greater speed and granularity than ever before. Modern AI tooling makes it possible to move seamlessly from spec to collection and from collection back to code — accelerating every stage of the API lifecycle. This 50-minute, hands-on bootcamp is designed to empower developers who want to build, test, and automate APIs and Model Context Protocol (MCP) servers using AI-powered workflows. You'll work in an environment where AI is always available, leveraging Postman Agent Mode to prototype, validate, and automate real, working API systems. You'll move quickly from concept to execution. In this session, you will: Prototype, test, and automate API workflows using Postman Agent Mode Generate environments, scripts, and documentation by describing what you need — not writing it Visualize and validate live API responses as rich, interactive dashboards Build, test, and validate an end-to-end integration test suite from a working API Understand how AI agents interact with APIs and MCP-enabled services in practice By the end of the bootcamp, you'll have hands-on experience with a complete, end-to-end workflow for building and validating AI-ready APIs using tools and techniques you can immediately apply in your own projects. We suggest bringing your laptop to make the most of this workshop" }, { "id": 3995, "day": "2026-09-30", "type": "Workshop", "start": "11:25", "end": "11:50", "track": [ "GreenIO" ], "room": [ "Doddington Forum 2" ], "title": "AI, ally or enemy of the planet?", "speakers": [ { "name": "Amael Parreaux-Ey", "title": "CEO and Co-founder", "company": "Resilio" } ], "facets": [ "accountability", "outside" ], "agent_readiness_dimensions": [], "why": "Environmental impact of AI as a governance tool; supply-chain transparency.", "abstract": "Today, no one really knows the energy consumption, carbon emissions or water consumption of AI models. This lack of transparency throughout the ICT supply chain poses a real problem for businesses that are exposed to supply-chain-related economic risks. Resilio is leading a collaborative study, involving academics, private and public organisations, and institutions, with the aim of raising awareness of the environmental impacts of AI, quantifying them, and turning this environmental information into a management and governance tool." }, { "id": 4048, "day": "2026-09-30", "type": "Talk", "start": "11:40", "end": "12:05", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "MCP, A2A, and what agentic protocols mean for API platform teams", "speakers": [ { "name": "Kaiwalya Koparkar", "title": "Platform Advocate", "company": "Gravitee.io" } ], "facets": [ "agent_readiness", "operational_transparency" ], "agent_readiness_dimensions": [ "mcp_server", "agent_card", "rate_limit_signal" ], "why": "Exposing APIs via MCP, A2A, token rate limits, data masking.", "abstract": "The Model Context Protocol (MCP) and agent-to-agent (A2A) communication patterns are rewriting assumptions about API design. This forward-looking session covers what these emerging protocols mean for API platform teams right now how to expose existing APIs through MCP servers, how to apply token rate limiting and data masking to LLM calls, and what \"API-first\" means in an agentic world." }, { "id": 4014, "day": "2026-09-30", "type": "Talk", "start": "11:40", "end": "12:05", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "Fostering AI skills and capabilities in large enterprises", "speakers": [ { "name": "Saheed Abiola Lasisi", "title": "Executive Director - Principal API Architect", "company": "JP Morgan Chase" } ], "facets": [ "accountability" ], "agent_readiness_dimensions": [], "why": "No abstract; AI skills inside a large enterprise.", "abstract": "" }, { "id": 4029, "day": "2026-09-30", "type": "Panel discussion", "start": "11:40", "end": "12:05", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "Operating Enterprise AI at Scale: From Autonomous Workflows to the Control Plane", "speakers": [ { "name": "David Wang", "title": "Head of Product, Marketing, and Customer Success", "company": "Tetrate" }, { "name": "Chris Falloon", "title": "Head, Portfolio Development", "company": "Dell Technologies Capital" } ], "facets": [ "agent_readiness", "accountability" ], "agent_readiness_dimensions": [ "delegated_identity", "agent_identity_declared" ], "why": "Identity, access, policy and visibility for autonomous workflows.", "abstract": "As AI moves beyond copilots toward autonomous systems that can reason, make decisions, and take action across enterprise workflows, organizations face a new set of operational challenges. Scaling these systems requires more than model performance—it demands an architecture that can securely connect AI workloads, establish identity and access, enforce policy, maintain visibility, and provide the right level of control without slowing innovation. Join Dell Technologies Capital and leaders building the next generation of enterprise AI infrastructure and applications for a discussion on what it takes to move autonomous AI into production at scale. From real-world AI workflows to the underlying control plane, the conversation will explore how enterprises are approaching secure connectivity, governance, runtime policy enforcement, observability, and trust as AI becomes increasingly embedded in mission-critical operations." }, { "id": 3996, "day": "2026-09-30", "type": "Workshop", "start": "11:50", "end": "12:15", "track": [ "GreenIO" ], "room": [ "Doddington Forum 2" ], "title": "Measured is not managed: what an IT estate looks like once the levers suggest themselves", "speakers": [ { "name": "James Hobbs", "title": "Head of Strategic Accounts & Alliances", "company": "Sopht" } ], "facets": [ "access_clarity", "outside" ], "agent_readiness_dimensions": [], "why": "FinOps + carbon levers across an estate.", "abstract": "FinOps, asset management, workplace, observability and carbon platforms each answer the question they were bought for, and none of them tells you where the largest gains sit across the whole estate. This workshop shows what it takes to see that in one place: a ranked list of optimisation levers, the cost and CO2 gain behind each, the complexity of doing it, and a simulation of the combined result before you commit." }, { "id": 4015, "day": "2026-09-30", "type": "Talk", "start": "12:05", "end": "12:30", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "The Mesh of API and AI Ecosystems", "speakers": [ { "name": "Nisu Thomas", "title": "Senior Enterprise Architect", "company": "Lloyds Banking Group" }, { "name": "Samyu Ayyar", "title": "Senior Enterprise Architect", "company": "Lloyds Banking Group" }, { "name": "Meghdeep Basu", "title": "Engineering Lead – API Platform", "company": "Lloyds Banking Group" } ], "facets": [ "regulatory", "agent_readiness" ], "agent_readiness_dimensions": [], "why": "API/AI mesh in banking; thin abstract.", "abstract": "To bring to life the interconnected link between API and AI to help unlock future capability for the Banking sector." }, { "id": 4030, "day": "2026-09-30", "type": "Talk", "start": "12:05", "end": "12:30", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "Computer Use via API: Giving Developers an Agent That Can See, Reason, and Act on Any Screen", "speakers": [ { "name": "Antonio Roberto Ventura", "title": "Founding Forward Deployed Engineer", "company": "H Company" } ], "facets": [ "agent_readiness", "access_clarity", "open_source" ], "agent_readiness_dimensions": [], "why": "Computer use is what agents do when there is no API. It is the counter-story to a Human-Only band.", "abstract": "This talk walks through what it actually takes to build and ship a computer-use model as an API. From the developer's perspective: how do you integrate an API whose output isn't a chat response but an autonomous action? How do you design an agentic loop, handle multi-step workflows across multiple applications, and manage state when your agent is navigating live enterprise software? From the model-builder's perspective: how did we architect a Mixture-of-Experts model (122B total, 10B active) to deliver flagship accuracy at API-friendly latency and cost? What tradeoffs did we make between our open-source 35B variant and the proprietary 122B — and how does a shared API surface let developers switch between them with a single parameter change?" }, { "id": 4049, "day": "2026-09-30", "type": "Workshop", "start": "12:05", "end": "12:55", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "Risk-Proof Your AI: The New API Governance Playbook", "speakers": [ { "name": "Markus Müller", "title": "Global CTO - API Management", "company": "Boomi" } ], "facets": [ "contract_governance", "accountability", "agent_readiness" ], "agent_readiness_dimensions": [], "why": "API governance as the risk control for agents.", "abstract": "Agentic AI brings greater autonomy—and greater risk. Strong API governance keeps agents secure, compliant & scalable, delivering value safely." }, { "id": 4042, "day": "2026-09-30", "type": "Workshop", "start": "12:05", "end": "13:00", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "Why you should already be exposing MCP? And how to do it in 5 easy steps", "speakers": [ { "name": "Dalli Bagdi", "title": "Integration Specialist", "company": "IBM" }, { "name": "Chris Phillips", "title": "Master Inventor", "company": "IBM" } ], "facets": [ "agent_readiness", "developer_ergonomics" ], "agent_readiness_dimensions": [ "mcp_server" ], "why": "Existing APIs -> MCP server in five steps.", "abstract": "Every major technology shift creates a change. REST transformed web applications. APIs accelerated digital transformation. AI agents are driving the next evolution, and MCP is rapidly becoming the standard way they interact with enterprise systems. If you're already exposing APIs, you're much closer than you think. In this session, we will demonstrate how to expose existing services as an MCP server in five easy steps using IBM Interact Gateway - without rewriting applications or modifying backend code. You'll leave with a practical understanding of MCP and a repeatable approach for making your existing APIs AI-ready." }, { "id": 3997, "day": "2026-09-30", "type": "Workshop", "start": "12:15", "end": "12:40", "track": [ "GreenIO" ], "room": [ "Doddington Forum 2" ], "title": "Free Greenpixie Data Now Available", "speakers": [ { "name": "Ben Price", "title": "Head of Operations", "company": "Greenpixie" }, { "name": "Stephen Old", "title": "Chief Operating Officer", "company": "Synyega" } ], "facets": [ "access_clarity", "discoverability", "outside" ], "agent_readiness_dimensions": [], "why": "Open carbon data for AWS/Azure/GCP compute.", "abstract": "The subject of cloud carbon emissions has long been a black box. Actionable, accurate data for unified cost and carbon optimisation is a dream that many engineers, FinOps practitioners, and IT leaders have given up on. Until now. Greenpixie's research team is releasing all its carbon data for AWS, Azure and GCP compute usage; so every forecast, optimisation, migration, modernisation and architecture design, can now feature carbon alongside cost, performance and security considerations. That's good news for those who believe technology and sustainability are better together. Hear how you can start using Greenpixie's industry-leading compute emissions data free today and what GreenOps experts like Synyega are already building on top of the free tier." }, { "id": 4025, "day": "2026-09-30", "type": "Talk", "start": "12:30", "end": "12:55", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "When Software Starts Improvising: How to Navigate the New World of Non-Deterministic Software Without Getting Burned.", "speakers": [ { "name": "Steven Willmott", "title": "former CEO of 3scale", "company": "CEO Safe Intelligence" } ], "facets": [ "agent_readiness", "accountability", "operational_transparency" ], "agent_readiness_dimensions": [ "dry_run_mode", "error_semantics" ], "why": "Verification and oversight for non-deterministic software.", "abstract": "Making an AI agent do something is only part of the job. Establishing that it does the right thing, reliably, can be much harder—creativity up often means reliability down. This talk covers a framework for thinking about software systems as they become infused with AI. We’ll explore how verification, validation, and testing can still help us. But it’s not all bad: creativity can also boost resilience when used well. Through practical examples, we’ll examine how to match architectural choices, testing, and operational oversight to the tasks: Which parts of the task should agents actually handle? What can we verify automatically? Where do we need broader evaluation and human judgment, and how do we handle the uncertainty that remains?" }, { "id": 3998, "day": "2026-09-30", "type": "Talk", "start": "12:40", "end": "13:00", "track": [ "GreenIO" ], "room": [ "Doddington Forum 2" ], "title": "Community Sponsors' Pitching Time", "speakers": [], "facets": [], "agent_readiness_dimensions": [], "why": "Sponsor pitches.", "abstract": "During Workshop Day, our community sponsors and some partners each get a few minutes on stage to share what they're building and why it matters for digital sustainability. The mic is open to all: whether you're a five-person startup or an established player, if you're working on this problem, you belong on this stage." }, { "id": 4054, "day": "2026-09-30", "type": "Talk", "start": "14:00", "end": "14:25", "track": [ "Apidays" ], "room": [ "Doddington Forum 2" ], "title": "APIs and AI Keynote: MCP vs. API Gateways: They’re Not Interchangeable", "speakers": [ { "name": "Goutam Verma", "title": "Software Development Engineer II", "company": "Expedia Group" } ], "facets": [ "agent_readiness", "operational_transparency" ], "agent_readiness_dimensions": [ "mcp_server", "protected_resource_metadata" ], "why": "MCP is stateful and session-based; gateways built for stateless HTTP do not fit.", "abstract": "Enterprises are rapidly adopting the Model Context Protocol (MCP) to expose internal systems and tools to AI agents, but many are running into a familiar architectural dilemma: how do you secure, govern, and operate MCP at scale? The instinctive answer is to reuse existing API gateways, but MCP is not an API, and treating it like one can lead to serious problems in production. This talk explores the fundamental differences between traditional APIs and MCP, and why those differences matter operationally. APIs are stateless and HTTP-driven, allowing gateways to enforce security, routing, and policy decisions using headers, paths, and methods. MCP, by contrast, is stateful, session-based, and built on JSON-RPC, with critical semantics living inside the request body and evolving over long-lived interactions. Streaming responses and server-initiated events further challenge the request response model most gateways assume. We’ll examine what actually happens when teams try to force MCP traffic through API gateways: brittle policy logic, limited observability, broken streaming enforcement, and growing coupling between infrastructure and application teams. From there, the talk walks through emerging MCP gateway patterns passthrough proxying, partial protocol awareness, brokering, and multiplexing and explains where each approach succeeds or fails. The goal is to help architects and platform teams make informed decisions as they expose MCP in production, avoid repeating hard-learned API mistakes, and design infrastructure that is ready for AI-native, agent-driven systems." }, { "id": 4017, "day": "2026-09-30", "type": "Panel discussion", "start": "14:00", "end": "14:25", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "100 AI Conversations in 20 Minutes: with SmartBear and Platformable", "speakers": [ { "name": "Joe Joyce", "title": "Solutions Architect", "company": "SmartBear" }, { "name": "Mark Boyd", "title": "Director", "company": "Platformable" } ], "facets": [ "developer_ergonomics", "contract_governance" ], "agent_readiness_dimensions": [], "why": "Patterns from hundreds of delivery-team conversations about AI.", "abstract": "Joe Joyce and Mark Boyd have spent the year having hundreds of conversations with API delivery teams about the opportunities and challenges presented by AI. Join them for a 20-minute tour through the insights, patterns, and lessons learned from these conversations." }, { "id": 4032, "day": "2026-09-30", "type": "Talk", "start": "14:00", "end": "14:25", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "Building generative media systems that actually work at scale", "speakers": [ { "name": "Wafae Bakkali", "title": "Generative AI Specialist, Tech Lead", "company": "Google" } ], "facets": [ "operational_transparency" ], "agent_readiness_dimensions": [], "why": "Reliability of generative media systems; weak API-surface signal.", "abstract": "Generative media models are maturing fast. The harder problem is building systems around them that work at scale across diverse inputs, shifting requirements, and the unforgiving expectations of production. This talk shares lessons learned from building generative media systems in enterprise environments: what breaks under real load, where the assumptions made in prototypes stop holding, and how to close the gap between a model's capabilities and a system's actual reliability. We'll cover evaluation strategies for media outputs, pipeline design for image and video generation and what it takes to ship end to end systems that hold up beyond the demo." }, { "id": 4050, "day": "2026-09-30", "type": "Workshop", "start": "14:00", "end": "14:50", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "Crazy things we've seen in the API world and how to avoid them", "speakers": [ { "name": "Ricky Moorhouse", "title": "Cloud Architect, API Connect", "company": "IBM" }, { "name": "Simon Kapadia", "title": "Chief Security Architect, API Connect", "company": "IBM" } ], "facets": [ "contract_governance", "contract_quality", "operational_transparency", "agent_readiness" ], "agent_readiness_dimensions": [], "why": "Anti-patterns: brittle design, inconsistent standards, security shortcuts.", "abstract": "After years of working with APIs, we’ve seen the same problems surface in different forms: brittle designs, inconsistent standards, governance gaps, security shortcuts, and architectures that become harder to evolve over time. In this session, we’ll unpack some of the most common API anti-patterns we’ve encountered, explain why they happen, and show practical ways to avoid them. You’ll learn how to design APIs that are easier to govern, secure, reuse, and adapt as requirements change, including as AI agents become a new class of API consumer." }, { "id": 4043, "day": "2026-09-30", "type": "Workshop", "start": "14:00", "end": "14:50", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "Let's Build, Secure, and Observe an AI Agent with an Agent Gateway", "speakers": [ { "name": "Gbadebo Bello", "title": "Senior Developer Relations Advocate", "company": "Postman" } ], "facets": [ "agent_readiness", "operational_transparency", "regulatory" ], "agent_readiness_dimensions": [ "auth_clarity", "delegated_identity" ], "why": "Centralised credentials, progressive tool disclosure, retries and observability in a banking demo.", "abstract": "AI agent workflows often begin as direct connections: the orchestration layer holds separate credentials, calls models and services independently, and manages tool access and reliability itself. In this hands-on workshop, you’ll take an existing Banking, Fraud, and Support agent workflow and route it through a single governed layer using Fabric Gateway. You’ll configure the gateway to connect the workflow’s LLM, MCP servers, and APIs; centralize upstream credentials; shape and progressively disclose the tool surface; and apply gateway-level controls for routing, retries, and observability. Using the accompanying comparison app, you’ll see the difference between the direct and gateway-mediated paths in real time. By the end, you’ll understand how to move a working agent orchestration from scattered service connections to a unified, governed execution path." }, { "id": 4018, "day": "2026-09-30", "type": "Talk", "start": "14:25", "end": "14:50", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "Govern APIs to Accelerate AI ROI", "speakers": [ { "name": "Markus Müller", "title": "Global CTO - API Management", "company": "Boomi" } ], "facets": [ "contract_governance", "accountability" ], "agent_readiness_dimensions": [], "why": "Governance under agents; MCP does not secure what is underneath.", "abstract": "AI agents run on APIs—but without governance, they bring data leaks, policy violations & instability. MCP speeds adoption; it doesn't secure what's underneath." }, { "id": 4055, "day": "2026-09-30", "type": "Talk", "start": "14:25", "end": "14:50", "track": [ "GenerationAI" ], "room": [ "Doddington Forum 2" ], "title": "Building MCPs in the era of the governance: MCP Ops", "speakers": [ { "name": "Jesus Vallejo Garcia", "title": "Head of APIs", "company": "cloudappi" } ], "facets": [ "contract_governance", "agent_readiness", "operational_transparency" ], "agent_readiness_dimensions": [ "mcp_server" ], "why": "MCP Ops: lifecycle, governance and security patterns for MCP servers.", "abstract": "In this session, we introduce the concept of MCP Ops. By drawing a direct parallel to the established world of API Ops, we will explore how to apply proven governance, lifecycle management, and security patterns to MCP development. Whether you are building custom servers or managing a fleet of AI connectors, it’s time to move beyond simple connectivity and start thinking about robust, governed operations." }, { "id": 4034, "day": "2026-09-30", "type": "Talk", "start": "14:50", "end": "15:15", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "When the Buyer is an Agent: How AI Intermediaries Disrupt eCommerce Acquisition", "speakers": [ { "name": "Katia Maevskaia", "title": "Commercial & Revenue Growth Leader", "company": "Snap Inc" } ], "facets": [ "agent_readiness", "discoverability" ], "agent_readiness_dimensions": [ "agentic_commerce" ], "why": "When the buyer is an agent, discovery-led retail breaks.", "abstract": "Agentic commerce is quietly breaking discovery-led retail. When an AI agent acts as an intermediary between the shopper and the store, the core mechanics of attribution, visual persuasion, and bidding auctions stop holding. Impressions are no longer served directly to human decision-makers, and persuasion strategies built for human psychology are priced into auctions that algorithms clear differently. In this session, Katia sets out what changes structurally for digital commerce and delivers an operational framework for what e-commerce, MarTech, and growth teams must instrument now to prepare for agent-mediated purchasing." }, { "id": 4019, "day": "2026-09-30", "type": "Panel discussion", "start": "14:50", "end": "15:15", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "Platform engineering panel: Invest in boundaries for speed and clarity", "speakers": [ { "name": "Jennifer Riggins", "title": "Tech Storyteller | Freelance Writer | Podcast Host | Tech Analyst", "company": "" }, { "name": "Abby Bangser", "title": "Founding Principal Engineer", "company": "Syntasso" }, { "name": "Molly Clarke", "title": "Senior cloud platform engineering manager", "company": "Hargreaves Lansdown" } ], "facets": [ "accountability", "operational_transparency", "agent_readiness" ], "agent_readiness_dimensions": [ "reversibility_documented" ], "why": "Ownership boundaries, incident management and reversible speed.", "abstract": "In this unrecorded panel for API architects, site reliability engineers, and platform engineers, we talk about the boundaries, guardrails, and communication processes that need to be in place to truly achieve sustainable, reversible speed. Join our discussion for how APIs define clear ownership across security and incident management, and through the management of engineering teams." }, { "id": 4056, "day": "2026-09-30", "type": "Talk", "start": "14:50", "end": "15:15", "track": [ "GenerationAI" ], "room": [ "Doddington Forum 2" ], "title": "Exposing Stochastic APIs", "speakers": [ { "name": "Sergey Konstantinov", "title": "Staff Software Engineer, Group Tech Lead", "company": "Bolt Technology (Estonia)" } ], "facets": [ "contract_quality", "agent_readiness", "upsert", "operational_transparency" ], "agent_readiness_dimensions": [ "idempotency", "error_semantics" ], "why": "What belongs in the contract and what to version when an LLM sits behind the API. What idempotency means when every call is a roll of the dice.", "abstract": "Put an LLM behind an API, and familiar rules start behaving strangely. \"\"200 OK\"\" can contain nonsense. Retrying the same request can produce a different answer. A model upgrade can break behavior without changing the schema. We'll explore how to design live APIs around these properties: what belongs in the contract, how to define semantic success, what exactly to version, and what retries and idempotency mean when every call is effectively another roll of the dice." }, { "id": 4051, "day": "2026-09-30", "type": "Workshop", "start": "14:50", "end": "15:40", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "How do you secure and govern every AI agent through one front door?", "speakers": [ { "name": "Kevin Peeters", "title": "Principal Pre-Sales Architect", "company": "Axway" } ], "facets": [ "agent_readiness", "contract_governance" ], "agent_readiness_dimensions": [ "well_known_catalog", "mcp_server", "auth_clarity" ], "why": "One governed front door for agent tools.", "abstract": "Axway Amplify Fusion and AI Gateway provide a single governed entry point for AI tools. A proven MCP Gateway pattern enables scalable agent access, discovery, and control without sacrificing oversight." }, { "id": 4044, "day": "2026-09-30", "type": "Workshop", "start": "14:50", "end": "15:40", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "A Better Way to Connect AI to Enterprise Data", "speakers": [ { "name": "Samuel Gaus", "title": "Principal Engineer and Team Lead", "company": "WunderGraph" }, { "name": "David Stutt", "title": "Senior Founding Engineer", "company": "WunderGraph" } ], "facets": [ "contract_governance", "discoverability", "contract_quality" ], "agent_readiness_dimensions": [], "why": "Agent-speed API sprawl; producer and consumer lifecycles that only meet in production.", "abstract": "Every developer on your team runs a coding agent now. Left alone, each one mints its own endpoint. Two agents pick up two similar tickets and ship two similar APIs, and nobody reviews either fast enough to catch it. Vibe coding is recreating microservice sprawl, at agent speed. We've all already agreed on the API lifecycle: producers plan, design, build, govern, deploy, evolve, retire; consumers discover, integrate, observe, adapt. We run those as two straight lines that only meet in production, usually during an incident. That was survivable when a new endpoint cost a sprint. It isn't when it costs twenty minutes. We can't fix it with more review because humans are already the bottleneck. Instead we need to put demand and governance inside the schema, where the agent has to deal with them to make progress. A consumer's unmet need becomes a typed, named gap in the graph rather than a stalled Slack thread. A change reaches its owners as a proposal to approve rather than a pull request to rubber-stamp. A second agent asking for the same field lands on the open proposal rather than minting a duplicate endpoint." }, { "id": 4068, "day": "2026-09-30", "type": "Talk", "start": "15:15", "end": "15:40", "track": [ "GenerationAI" ], "room": [ "Doddington Forum 2" ], "title": "Do Tools Still Matter? MCP Tool Design in the Age of Code Mode", "speakers": [ { "name": "Ruben Casas", "title": "Senior Engineer", "company": "Postman" } ], "facets": [ "agent_readiness", "contract_quality" ], "agent_readiness_dimensions": [ "mcp_server", "operation_distinctiveness" ], "why": "Tool design vs Code Mode; bloated tool surfaces.", "abstract": "The first wave of MCP adoption was defined by abundance. Every API endpoint became a tool, every workflow became a collection of tool calls, and many servers exposed hundreds of operations directly to agents. It worked, but at a cost: bloated context windows, slower tool selection, and increasingly unreliable agent behaviour. But then Code Mode appeared! With approaches Code Mode agents can search an API surface, generate code, and interact with services directly. If models can consume entire APIs and write code to acomplish a task, does tool design still matter? In this talk, we'll compare three approaches to agent tooling: curated workflow-oriented tools, tool discovery and search, and Code Mode's search-and-execute model. Using real-world examples, benchmarks, and evals, we'll explore the trade-offs in reliability, latency, token usage, and task success. Attendees will leave with a practical framework for deciding when to consolidate tools, when to expose APIs, and whether the future of MCP is better tools or fewer tools altogether." }, { "id": 4120, "day": "2026-09-30", "type": "Talk", "start": "15:15", "end": "15:40", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "APIs in the Age of AI Built for Humans, Now Consumed by Agents", "speakers": [ { "name": "Asanka Abeysinghe", "title": "CTO", "company": "WSO2" } ], "facets": [ "agent_readiness", "developer_ergonomics" ], "agent_readiness_dimensions": [], "why": "APIs built for humans, consumed by agents. No abstract.", "abstract": "" }, { "id": 4035, "day": "2026-09-30", "type": "Talk", "start": "15:15", "end": "15:40", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "The question of spending AI tokens is becoming a hot topic, what is the future of tokenmaxxing and how to optimize token spend.", "speakers": [ { "name": "Artem Matinian", "title": "CEO", "company": "Strukt.ai" } ], "facets": [ "access_clarity" ], "agent_readiness_dimensions": [], "why": "Token spend: pricing and finops.", "abstract": "Corporations, small startups are spending millions on AI credit usage. What does the future hold is uncertain, developers and engineers are already dependants on OpenAI, Anthropic and other AI tools in order to create and manage code. But does the average manager knows what the credit spend is being used for? We will talk about what we think can be the future of startups, what happens if the flagship models are going to make a significant price increase and more." }, { "id": 4021, "day": "2026-09-30", "type": "Talk", "start": "16:00", "end": "16:25", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "Syncing the Shuffle: Architecting Real-Time APIs to Drive Complex Game Animations in NYT Crossplay", "speakers": [ { "name": "Shafik Quoraishee", "title": "Senior Game Developer", "company": "The New York Times" } ], "facets": [ "contract_quality", "operational_transparency" ], "agent_readiness_dimensions": [ "event_surface_described" ], "why": "Real-time APIs for game animation; mostly internal.", "abstract": "The New York Times has long defined the daily puzzle habit, but the launch of Crossplay—our real-time multiplayer word game—required a massive leap in how our systems communicate. Building complex board interactions like tile shuffling, zoom mechanics, and fluid drag-and-drop physics without a traditional game engine is a steep UI challenge. But keeping those animations instantaneous, perfectly synchronized, and seamless across millions of devices? That is an API and data architecture challenge. In this talk, we’ll pull back the curtain on how the engineering team architected high-performance, real-time APIs to drive world-class native game mechanics. We will explore why we abandoned standard WebSockets for a broadcast-based Fanout system, how we structure payloads to trigger complex tile animations, and the blueprint for making network-driven micro-interactions feel completely frictionless to the end user." }, { "id": 4036, "day": "2026-09-30", "type": "Talk", "start": "16:00", "end": "16:25", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "AgentGuard: Securing Inter-Agent Messages", "speakers": [ { "name": "Umme Nizba", "title": "Software Engineer", "company": "" } ], "facets": [ "agent_readiness", "open_source", "accountability", "operational_transparency" ], "agent_readiness_dimensions": [ "consent_identity", "agent_identity_declared" ], "why": "Open-source middleware: message attestation, capability manifests, audit logs.", "abstract": "Multi-agent systems treat APIs and MCP tools as the “hands” of AI — but most security still stops at the user prompt. Attacks arrive through fetched web content, forged inter-agent messages, and tool returns; the downstream agent that executes the call often holds the credentials. This session presents AgentGuard, open-source middleware for inspecting and governing traffic between agents: rules and optional ML scoring, Ed25519 message attestation, YAML capability manifests (which APIs each agent may call), and tamper-evident audit logs. A live IntelBrief demo shows a three-agent LangGraph pipeline where poisoned web content triggers an unauthorised external publish in the unsecured run and is blocked in the secured run. For teams building agentic finance and embedded banking, this is the gap between “the API is authenticated” and “this specific agent was allowed to invoke this specific endpoint with this payload.”" }, { "id": 4069, "day": "2026-09-30", "type": "Talk", "start": "16:00", "end": "16:25", "track": [ "Apidays" ], "room": [ "Doddington Forum 2" ], "title": "The SaaS Hangover: From SaaS Dependency to Digital Sovereignty", "speakers": [ { "name": "Andrew Murray", "title": "Chief Customer Officer", "company": "Tyk" } ], "facets": [ "access_clarity", "regulatory", "open_source" ], "agent_readiness_dimensions": [], "why": "SaaS dependency, jurisdiction and digital sovereignty.", "abstract": "The SaaS Hangover: From SaaS Dependency to Digital Sovereignty explores how the convenience of SaaS has created new forms of dependency on vendors, cloud platforms and jurisdictions. As digital sovereignty rises up the agenda, this talk looks at where organisations have lost control, how AI is deepening those dependencies, and how to make more deliberate technology choices without giving up the benefits of modern cloud software." }, { "id": 4045, "day": "2026-09-30", "type": "Workshop", "start": "16:00", "end": "16:50", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "Federated API Intelligence in Action: A Hands-On Demo with Treblle", "speakers": [ { "name": "Vedran Cindrić", "title": "Founder & CEO", "company": "Treblle" } ], "facets": [ "operational_transparency", "discoverability", "contract_governance" ], "agent_readiness_dimensions": [], "why": "API observability and inventory.", "abstract": "Every company is becoming an API company, and AI is speeding that up. In this interactive demo, Treblle's founder and CEO shows how Treblle gives your teams full visibility into the APIs behind your digital transformation. You'll see how it cuts manual work, reduces risk, and gets you ready for AI, so you can move faster." }, { "id": 4052, "day": "2026-09-30", "type": "Workshop", "start": "16:00", "end": "16:50", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "When API Tokens Become Attack Vectors", "speakers": [ { "name": "Sachin Thombre", "title": "AppSec Sales Specialist - EMEA", "company": "Thales" } ], "facets": [ "agent_readiness", "operational_transparency" ], "agent_readiness_dimensions": [ "auth_scheme_strength", "auth_clarity" ], "why": "Tokens as attack vectors, which is exactly bound vs bearer.", "abstract": "APIs are only as secure as their tokens. Explore authentication risks, exposed credentials and attack patterns—and learn how to strengthen API security before attackers exploit them." }, { "id": 4022, "day": "2026-09-30", "type": "Talk", "start": "16:25", "end": "16:50", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "Google Talk", "speakers": [], "facets": [], "agent_readiness_dimensions": [], "why": "Google talk; no abstract.", "abstract": "" }, { "id": 4070, "day": "2026-09-30", "type": "Talk", "start": "16:25", "end": "16:50", "track": [ "Apidays" ], "room": [ "Doddington Forum 2" ], "title": "Modernize Your Middleware for the Agentic Era", "speakers": [ { "name": "Akshata Sawant", "title": "Lead Developer Advocate", "company": "Salesforce" } ], "facets": [ "contract_governance", "agent_readiness", "developer_ergonomics" ], "agent_readiness_dimensions": [ "mcp_server" ], "why": "Middleware modernisation for agents.", "abstract": "Your middleware was built for a different world. Today's enterprises need integration that does more than connect systems it has to power agents, enforce governance, and scale across an AI-driven architecture. In this session, discover how MuleSoft helps you evolve beyond aging middleware like Spring Boot and legacy ESBs to build the connectivity backbone your agentic enterprise demands. Learn how to reuse existing investments while unlocking new capabilities: real-time API management, agent orchestration, and composable integrations that connect any system to any AI. See how organizations are accelerating their modernization journey with MuleSoft as the bridge between where they are today and where agentic AI needs them to be. Walk away with a practical framework for moving from legacy middleware to a future-ready, governed integration platform without disrupting what's already working." }, { "id": 4037, "day": "2026-09-30", "type": "Talk", "start": "16:25", "end": "16:50", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "End to End encrypted Agentic communication control-place for long living agents", "speakers": [ { "name": "Farshid Ashouri", "title": "Head of Data Engineering @ JPMorganChase", "company": "JPMorganChase" } ], "facets": [ "agent_readiness", "operational_transparency" ], "agent_readiness_dimensions": [ "agent_identity_declared", "agent_card" ], "why": "Addresses and a bus for long-lived agents.", "abstract": "Right now you are the message bus. You copy an error out of one agent and paste it into another, carry answers back, and hold in your head who is still waiting on what. A communication bus with scheduler and wakeup call via SMTP gives them addresses so they can do that part without you." }, { "id": 4147, "day": "2026-09-30", "type": "Talk", "start": "16:25", "end": "16:50", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "Building Enterprise-ready Agents with an AI Gateway", "speakers": [ { "name": "Shruti Naik", "title": "Engineering Lead – Kubernetes", "company": "Lloyds Banking Group" }, { "name": "Nigel Walters", "title": "Customer Engineer", "company": "Google Cloud" }, { "name": "Bradley Cummings", "title": "Apigee Product Owner", "company": "Lloyds" } ], "facets": [ "agent_readiness", "contract_governance", "regulatory" ], "agent_readiness_dimensions": [], "why": "Agentic sprawl behind an AI gateway at a bank.", "abstract": "Understanding the demands of agentic sprawl and the risks that need to be addressed are key to any organisation moving forward." }, { "id": 4154, "day": "2026-09-30", "type": "Talk", "start": "16:50", "end": "17:15", "track": [ "Apidays" ], "room": [ "Doddington Forum 2" ], "title": "Making sense of application infrastructure in AI times", "speakers": [ { "name": "Paolo Malinverno", "title": "Chief strategy officer", "company": "Sensedia" } ], "facets": [ "contract_governance", "agent_readiness", "accountability" ], "agent_readiness_dimensions": [], "why": "API, event, MCP, LLM and context management stacked up; 'just enough governance' as an illusion.", "abstract": "brief history (not more than 10 m) of how the various components came along (API management, Events management; MCP management; LLM management, Context Management, but also identity, MCP servers and AI gateway) just dumping all AI stuff on top of that mess won't work if you think designing AI agents on top of that is complicated, wait to see how AI Agents (they are new users) proliferation will make unprecedented demands on it how to make sense of the mess: the history is repeating itself, just enough governance sadly is going to prove an illusion yet again the disappearance of maintenance (and a lot of testing) a few myths to get out of the way (e.g. you don't need to do as many agents as possible, or single X is a myth, multi X is a reality) Model routing, MCP gateways, Agent gateways bring a lot of promises You don't have to use AI for everything, because it doesn't do everything: use human and AI agent teams, design AI Agents right from the beginning and protect your application stack" }, { "id": 4112, "day": "2026-09-30", "type": "Talk", "start": "16:50", "end": "17:15", "track": [ "Apidays" ], "room": [ "Grand Hall 1" ], "title": "Agentic Engineering at Scale: Strategy, Problems, and Where It's Going", "speakers": [ { "name": "Rajani Rao", "title": "Founder of Women Coding Community", "company": "Principal Technologist at AVEVA" } ], "facets": [ "accountability", "regulatory" ], "agent_readiness_dimensions": [], "why": "Agentic engineering against critical-infrastructure legacy code; internal.", "abstract": "Agentic engineering promises faster delivery. Industry data (DORA's 2025 report) shows a paradox: teams adopting AI tools often report feeling faster, while delivery metrics don't always follow. This talk is a practical account of scaling agentic engineering across AVEVA's engineering organization, a company built on decades of industrial automation software, legacy systems, and the kind of complexity that comes with running critical infrastructure for energy, utilities, and manufacturing. That context matters. Agentic engineering is easy to demo on a greenfield app. It's a different problem entirely against legacy codebases, long-lived architecture decisions, and systems where reliability isn't optional. I'll cover the strategy we built for that environment, not just tooling, but people, culture, platform, and governance. Then the problems: some of which we have solved and some we're still working through, and where it's all going. I'll also share what we built to close the gap between \"\"feels faster\"\" and \"\"is faster,\"\" and how it changes what gets measured, what gets reviewed, and where human judgment still has to sit. It's a working account for engineering leaders and platform teams dealing with agentic engineering against real complexity, not a clean slate." }, { "id": 4046, "day": "2026-09-30", "type": "Talk", "start": "16:50", "end": "17:15", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "What is a Customer Context API?", "speakers": [ { "name": "Gareth Faull", "title": "Senior Product Manager - Platform API Gateway & Access", "company": "London Stock Exchange Group" } ], "facets": [ "access_clarity", "operational_transparency", "agent_readiness", "regulatory" ], "agent_readiness_dimensions": [ "rate_limit_signal", "auth_clarity" ], "why": "Entitlements, scopes, rate limits, contracts, incidents and deprecations as an API: the Kin Score facets made customer-specific.", "abstract": "Most API platforms are built around a catalogue: a list of APIs, documentation, onboarding flows, sandboxes and access requests. That works reasonably well when every customer is treated as a generic developer reading generic docs. But real API customers are not generic. They have specific products, service accounts, entitlements, environments, scopes, usage patterns, rate limits, contracts, incidents, deprecations and risks. This talk argues for a missing layer in API platforms: the Customer Context API. Not another business API, but a way for customers, internal teams and eventually AI agents to ask: what can this customer use, what are they using, what constraints apply, what has changed, what is at risk, and what should happen next? Documentation tells you what an API is. Context tells you what it is to you. If API platforms are going to support increasingly automated, regulated and customer-specific integration, they need to expose that context as a product capability rather than leaving it scattered across IAM, gateways, billing, CRM, support tickets and tribal knowledge." }, { "id": 4038, "day": "2026-09-30", "type": "Talk", "start": "16:50", "end": "17:15", "track": [ "GenerationAI" ], "room": [ "Grand Hall 2" ], "title": "Who Let the Agent In? Securing MCP Servers in Production", "speakers": [ { "name": "Prachi Jamdade", "title": "Developer Advocate", "company": "Gravitee" } ], "facets": [ "agent_readiness" ], "agent_readiness_dimensions": [ "protected_resource_metadata", "dynamic_client_registration", "auth_clarity", "delegated_identity" ], "why": "MCP-spec-compliant authn/authz in production.", "abstract": "The Model Context Protocol (MCP) is reshaping how agents interact with tools and APIs, but building MCP servers that are secure, governed, and production-ready remains a challenge. Many teams want to expose powerful capabilities through MCP, yet struggle to implement authentication and authorization that follow the MCP specification while staying flexible for real-world use cases. This talk explores how to implement MCP-compliant authentication and robust authorization models without adding unnecessary complexity. We begin with a clear overview of how MCP handles identity and access, followed by a walkthrough of a minimal MCP server implementation. Once the foundation is in place, we introduce standards-aligned authentication and demonstrate how to design fine-grained, contextual authorization using OpenFGA. We then connect these concepts to real-world API governance scenarios, where multiple agents and tools require controlled access to domain-specific operations. As agent-driven architectures evolve, ensuring secure and structured access becomes essential. Finally, we explore how platforms like Gravitee can offload authentication and authorization entirely, including FGA-style access control, through a fully managed, no-code approach. This allows developers to focus on building MCP servers instead of maintaining complex security layers." }, { "id": 4053, "day": "2026-09-30", "type": "Workshop", "start": "16:50", "end": "17:40", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "Apigee Ai Gateway labs on Google Cloud", "speakers": [ { "name": "Saravanakumar Rajagopal", "title": "Solutions Architect", "company": "Google Cloud" }, { "name": "Nigel Walters", "title": "Customer Engineer", "company": "Google Cloud" } ], "facets": [ "agent_readiness", "contract_governance" ], "agent_readiness_dimensions": [], "why": "Apigee AI Gateway lab.", "abstract": "This lab will guide you through an AI Gateway on Google leveraging Apigee, Model Garden, Cloud Networking and Security Services to secure & govern AI traffic to and from models, tools and agents." }, { "id": 4171, "day": "2026-09-30", "type": "Talk", "start": "17:15", "end": "17:40", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "Build once, reuse everywhere: scaling GenAI use cases across the enterprise", "speakers": [ { "name": "Kateryna Korotieieva", "title": "AI Principal Product Manager", "company": "GlobalLogic, Hitachi group" } ], "facets": [ "contract_governance", "discoverability", "accountability" ], "agent_readiness_dimensions": [], "why": "Reuse across GenAI use cases instead of rebuilding each one.", "abstract": "Most enterprises start their GenAI journey one use case at a time. It makes sense at first: a team picks a problem, builds a solution, and proves value. But by the fifth or tenth use case, a pattern appears across the industry. Teams are solving the same problems again and again: connecting to data, retrieving the right information, meeting security and governance requirements, and getting sign-off to go live. Each new use case costs almost as much as the first, and scaling stalls. This talk shares a practical approach to breaking that cycle. Drawing on my experience delivering 15+ GenAI use cases in a global enterprise, I'll show how to spot what's worth reusing and what isn't, how to design shared capabilities that different teams will actually adopt, and why governance built in from the start speeds delivery up rather than slowing it down. This matters because the next stage of enterprise GenAI isn't more pilots. It's making each new use case faster, cheaper and safer to deliver. Attendees will leave with a simple framework for deciding what to build once, and how to make it reusable everywhere." }, { "id": 4159, "day": "2026-09-30", "type": "Talk", "start": "17:15", "end": "17:40", "track": [ "Apidays" ], "room": [ "Doddington Forum 2" ], "title": "Your Next Developer Might Be a Robot: DevRel in the AI Era", "speakers": [ { "name": "Paul Ardeleanu", "title": "Principal Architect", "company": "Vonage" } ], "facets": [ "developer_ergonomics", "discoverability", "agent_readiness" ], "agent_readiness_dimensions": [ "agent_skills", "well_known_catalog" ], "why": "DevRel when the reader is an agent parsing docs and samples at 3AM.", "abstract": "Developer Relations has always been about meeting developers where they are, but what happens when \"developers\" aren't just humans anymore? AI agents are rapidly becoming first-class consumers of our APIs, documentation, and tooling, and they don't attend meetups or watch conference talks but are rather parsing your public documentation and sample repositories at 3AM, making decisions based on what they discover. It's happening now and it fundamentally changes what Developer Relations needs to be. This talk explores how DevRel must evolve when the audience has rapidly expanded to include two new categories of users: humans who build with the help of AI-enabled tools, and AI that operates autonomously. We'll dig into the critical questions: is the documentation agent-readable? Are the APIs designed for non-human consumers? What does \"\"developer experience\"\" mean when the developer never sleeps and has no patience for ambiguity? We'll also address the new challenges head-on: trust gets harder when agents act on behalf of users, support gets messier, and traditional success metrics start to break down. The upside is proportionally substantial: DevRel teams equipped with AI tools can produce more content, reach more developers, and respond faster than ever before, multiplying their impact. Whether you're in DevRel, building developer tools, or just curious about where this is all heading, this session will give you a practical framework for thriving when your newest power users don't have a physical presence. So far…" }, { "id": 4024, "day": "2026-09-30", "type": "Talk", "start": "17:15", "end": "17:40", "track": [ "Apidays" ], "room": [ "Grand Hall 1", "Doddington Forum 1" ], "title": "Challenges and Opportunities in API Utilisation and Governance Across Global Governments: A Study of Best Practices and Success Stories", "speakers": [ { "name": "Dr Ravinder Singh", "title": "Head of Digital and Systems Team", "company": "Cabinet Office-GCF" } ], "facets": [ "regulatory", "contract_governance", "discoverability" ], "agent_readiness_dimensions": [], "why": "API governance across governments. No abstract.", "abstract": "" }, { "id": 3911, "day": "2026-10-01", "type": "Talk", "start": "08:45", "end": "10:00", "track": [ "Apidays", "Women in API" ], "room": [ "Doddington Forum 1" ], "title": "Women in APIs breakfast", "speakers": [ { "name": "Claire Barrett", "title": "Founder", "company": "APIsFirst" }, { "name": "Akshata Sawant", "title": "Lead Developer Advocate", "company": "Salesforce" }, { "name": "Sophie Sarin", "title": "Business Application Platform Lead", "company": "Google" } ], "facets": [], "agent_readiness_dimensions": [], "why": "Women in APIs breakfast.", "abstract": "Step into a supportive space where connections thrive Introducing the latest edition of our popular in-person GET /Connecting meet-up series, as part of the upcoming apidays London conference. When is it? We'll be connecting over breakfast on 01 October 2026 from 9.00am to 10.00am BST. Why join? You'll meet like-minded individuals in a friendly environment. You'll make new connections with other conference participants: attendees, speakers and panellists. You'll have meaningful interactions with them up front, setting you up for a richer overall conference experience. You'll get breakfast! Other attendees include fellow WIA members and allies from the wider community of API professionals and partners. Who's invited? Everyone is welcome to our GET /Connecting meetups, regardless of gender. You must, however, be registered to the apidays London 2026 conference. What is Women in APIs? Women in APIs (WIAs) provides a platform for a diverse global community of tech and consulting experts, particularly within the API industry. We are about helping people to be seen and heard in the public domain. We have more than 1,100 members from 64 countries. You can find out more about who we are and what we do at www.womeninapis.com ." }, { "id": 3981, "day": "2026-10-01", "type": "Opening remarks", "start": "09:30", "end": "09:45", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "Opening remarks", "speakers": [ { "name": "Gael Duez", "title": "Founder & Host", "company": "Green IO" } ], "facets": [ "outside" ], "agent_readiness_dimensions": [], "why": "GreenIO opening: Follow the energy.", "abstract": "Somewhere in the last three years, writing code became an energy decision. Not figuratively. A model choice is a power draw. A retry policy is a power draw. A default context window, shipped once and forgotten, is a power draw repeated millions of times a day. None of that used to be ours to decide, and most of us missed the handover. That is Follow the energy. Over two days at Convene Sancroft we trace one watt from the grid, where a connection queue decides what gets built, through the data centre, into the codebase, up to the board. Everyone here holds a piece of it: sustainability experts working on procurement and decarbonisation plans, Tech practitioners working on GreenOps and AI right-sizing. Some of us will flatly disagree this week, which is rather the point of a room. Responsible technologists do not need another warning. We need numbers, levers and someone who has tried it. Thank you to the speakers, partners and volunteers who brought them. Welcome back." }, { "id": 4020, "day": "2026-10-01", "type": "Talk", "start": "09:30", "end": "09:55", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "Who Governs the Agents? Rethinking AI Gateways in Agentic Architectures", "speakers": [ { "name": "Matthias Biehl", "title": "API Evangelist", "company": "IBM" } ], "facets": [ "agent_readiness", "contract_governance", "accountability" ], "agent_readiness_dimensions": [], "why": "AI gateway sprawl across API, model, LLM and A2A layers.", "abstract": "The promise of the Agentic Enterprise depends on one critical assumption: we can trust agents to do the right things. But AI agents do not operate in isolation. They are deeply embedded in enterprise landscapes—calling APIs and MCP servers, invoking LLMs and AI models, and collaborating with other specialized agents through A2A patterns. Every one of these interactions introduces both value and risk. To manage that risk, enterprises have introduced governance at every layer—API gateways, AI gateways, model gateways, LLM gateways, A2A gateways, agent frameworks, and security controls. The result? AI gateway sprawl. Practitioners now face an increasingly fragmented control plane, with overlapping gateways, inconsistent policies, and limited end‑to‑end visibility into agent behavior. This session explores the emerging role of AI Gateways as the governance backbone of the Agentic Enterprise. Attendees will leave with a clearer mental model for Agentic Enterprise Architecture, how to govern agents without slowing innovation, and how to tame and avoid AI gateway sprawl." }, { "id": 4013, "day": "2026-10-01", "type": "Talk", "start": "09:30", "end": "09:55", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "From APIs to MCP: Enterprise Governance, Registry, and Controls", "speakers": [ { "name": "Stefan Mesquita", "title": "Platform Experience Lead at Deutsche Bank", "company": "Deutsche Bank" } ], "facets": [ "discoverability", "agent_readiness", "contract_governance", "regulatory" ], "agent_readiness_dimensions": [ "mcp_server", "well_known_catalog" ], "why": "One registry for APIs and MCP servers at a bank.", "abstract": "APIs remain the foundation of enterprise integration, but AI agents are changing how capabilities are discovered and consumed. MCP standardizes tool discovery and invocation, enabling agents to interact with enterprise systems through reusable tools. The challenge is not enabling MCP. The challenge is governing it at scale. In this talk, we will discuss how to operationalize MCP in an enterprise by leveraging existing API Management capabilities instead of creating a parallel governance stack. We will cover how to build a shared registry for APIs and MCP servers that acts as the discovery layer for modern agents, and we will walk through the governance controls that matter in practice. Attendees will leave with a practical blueprint for implementing MCP safely and repeatably at enterprise scale, making MCP adoption scalable, observable, and compliant." }, { "id": 4108, "day": "2026-10-01", "type": "Talk", "start": "09:30", "end": "09:55", "track": [ "OpenAPI" ], "room": [ "Doddington Forum 2" ], "title": "Beyond OpenAPI: API Workflows That AI Agents Can Actually Execute", "speakers": [ { "name": "Frank Kilcommins", "title": "Head of Enterprise Architecture", "company": "Jentic" } ], "facets": [ "contract_quality", "agent_readiness", "developer_ergonomics" ], "agent_readiness_dimensions": [ "event_surface_described" ], "why": "Arazzo 1.1: executable workflows across OpenAPI and AsyncAPI.", "abstract": "AI agents can call individual API endpoints, but multi-step workflows? That's where they struggle. OpenAPI describes what each endpoint does, but not how to orchestrate them into business processes. Agents are left to infer sequencing, handle conditional logic, and manage state across calls—burning tokens and producing unpredictable results. The Arazzo Specification solves this by providing machine-readable workflow descriptions that agents can execute directly. And Arazzo 1.1 takes it further, extending support from OpenAPI-based REST services to AsyncAPI-defined event systems, enabling true workflow orchestration across synchronous and asynchronous boundaries. This talk explores what's new in Arazzo 1.1: AsyncAPI support for event-driven workflows, enhanced selector support using JSONPath and XPath for precise data extraction, and chained workflow composition for reusable process libraries. These capabilities transform how agents interact with APIs—from inferring workflows to executing defined processes. Whether you're building automation across distributed systems, exposing use-case-oriented capabilities to AI agents, or designing MCP servers with governed capability access, Arazzo 1.1 provides the declarative fabric for predictable, testable, agent-executable workflows. If OpenAPI tells agents what they can call, Arazzo tells them what they should do—and how to do it reliably." }, { "id": 3999, "day": "2026-10-01", "type": "Keynote", "start": "09:45", "end": "10:20", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "Can offshore wind power the AI boom?", "speakers": [ { "name": "Susan Gourvenec", "title": "Professor of Offshore Geotechnical Engineering", "company": "University of Southampton" } ], "facets": [ "outside" ], "agent_readiness_dimensions": [], "why": "Offshore wind and AI demand.", "abstract": "Deployment of offshore wind has accelerated over the last two decades but needs to increase pace and scale of deployment to meet mid-century net zero targets. I will explore how the AI boom is adding to the demand for decarbonised electricity and the opportunities for the offshore wind sector to respond through technology and policy for responsible, sustainable solutions. As two relatively new kids on the block, I will set out the supply and demand of offshore wind and AI in the broader societal and energy context. I will debunk some myths and shine a light on the risk of inaction for us to decarbonise and to ensure we stay on the leading edge of both offshore wind and AI." }, { "id": 4089, "day": "2026-10-01", "type": "Talk", "start": "09:55", "end": "10:20", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "How APIs become a revenue engine", "speakers": [ { "name": "Olga Kostina", "title": "Product Manager Payments API", "company": "Adyen" } ], "facets": [ "contract_quality", "operational_transparency", "access_clarity", "regulatory" ], "agent_readiness_dimensions": [], "why": "Optional vs mandatory fields, validation, versioning and backward compatibility as revenue levers.", "abstract": "In payments, API design affects conversion, authorization, fraud performance, and how well ML can optimise outcomes. Much of that comes down to data quality: collecting the right information, at the right moment, with the right level of API validation. This talk explores the tradeoffs API teams face when trying to improve business performance through better API design. How do we balance ease of integration with the need for stronger data quality? When should a field stay optional, and when does making it mandatory unlock measurable value? How do legacy integrations, versioning constraints, and backward compatibility slow down progress? Using a practical product and platform perspective, this session shows how API teams can move beyond maintaining contracts and start influencing revenue outcomes. It matters because the quality of an API is not just a developer experience question. It can directly impact growth, optimisation, and long term platform evolution." }, { "id": 4109, "day": "2026-10-01", "type": "Talk", "start": "09:55", "end": "10:20", "track": [ "OpenAPI" ], "room": [ "Doddington Forum 2" ], "title": "OpenAPI Extensions Are Not a Mess, They Are a Map", "speakers": [ { "name": "Kin Lane", "title": "Founder", "company": "API Evangelist" } ], "facets": [ "contract_quality", "contract_governance", "agent_readiness", "discoverability" ], "agent_readiness_dimensions": [ "extensions_in_served_contract" ], "why": "Kin's talk: 1,637 x- extensions across 15,722 OpenAPIs from 7,012 providers.", "abstract": "Five years ago I looked at about fifty OpenAPI extensions from fourteen providers and concluded that providers were quietly bending OpenAPI to carry operational details the specification was never designed to hold. This year I pointed a much bigger lens at the same question using my APIs.io search engine — every OpenAPI in my catalog, 15,722 documents from 7,012 providers I found 1,637 distinct 'x-' extensions. delivering on a variety of operational concerns. Some you would guess: SDK generation, documentation, mocking and testing. Others are there because the specification has no word for them at all — provenance (where did this document come from), visibility (who is allowed to see this operation), capability declaration (what can this endpoint actually do), change management, monetization (what does a call cost), legal, privacy. Not one of those is a fringe concern, and not one of them has a home. The biggest job by far is saying something about data that OpenAPI and JSON Schema could not, and inside it sit nine more distinct problems, from type systems to identity and relationships to units and formatting. A regulator needing to say 'this is a CDS PositiveInteger', a payments company needing to say 'this is money', and a date library all hit the same wall separately and each invented their own way through it. Like most of my work with API, I do not see this as a mess we should clean this up. Extensions are why OpenAPI has survived contact with reality. It is that eighteen distinct jobs is not a specification gap you patch — it is the wrong container. Almost none of what these extensions carry is the API contract. It is operational metadata about the contract, and it belongs alongside it, in Overlays and APIs.json, rather than crammed inside an OpenAPI." }, { "id": 4121, "day": "2026-10-01", "type": "Talk", "start": "09:55", "end": "10:20", "track": [ "Apidays" ], "room": [], "title": "Orchestrating Integration Combining AgenticAI and APIs", "speakers": [ { "name": "Radhouane Jrad", "title": "Enterprise Integration Architect", "company": "DLA Piper" } ], "facets": [ "agent_readiness" ], "agent_readiness_dimensions": [], "why": "Combining agents and APIs; thin abstract.", "abstract": "Agentic AI is a hot topic, and APIs are a mature concept. Using one or the other returns a lower business value than when combining both." }, { "id": 4146, "day": "2026-10-01", "type": "Talk", "start": "10:20", "end": "10:45", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "The Missing Architect: Telco APIs are stuck between everyone’s lane. Who should bridge the gap?", "speakers": [ { "name": "Nauman Saeed Anwar", "title": "Senior Go to Market PM", "company": "GSMA" } ], "facets": [ "regulatory", "contract_governance", "access_clarity", "discoverability" ], "agent_readiness_dimensions": [], "why": "CAMARA network APIs are live yet adoption lags. The gap is across remits, not technology.", "abstract": "The APIs are live. The demand is documented. So why does a launch still take quarters? Network APIs are commercially available. Operators make them available via CAMARA standard, technology partners package them into products, and enterprises are publishing their requirements. Yet in many markets adoption moves more slowly than demand, and the constraint is rarely technology or appetite. A scam crosses telecom, privacy and payments remits in minutes. The response usually stays within one. That gap is the whole problem in miniature. Operators see the network. Technology partners see their own stack. Telecom regulators apply telecom law, data protection authorities apply privacy law, financial regulators apply payments law. Each view is correct. None is the whole picture, and taking on the whole picture can look like antitrust exposure or a cross-border data question, so the work waits. This is a practitioner view from Open Gateway go-to-market activity across European, Middle Eastern and African markets: what shapes the timeline from published standard to first commercial call, why an approval in one market rarely carries to the next, and what the faster launches have had in common. United Against Scams showed what becomes possible when operators, banks and regulators align on a single use case in a protected forum. That way of working now has to extend to Number Verification, Quality on Demand and other non-fraud prevention use cases. The session sets out what brings the right parties to the table early, even when the case for moving quickly is less immediate, and what practitioners and enterprises can do to make it happen faster in their own markets and segments." }, { "id": 4099, "day": "2026-10-01", "type": "Talk", "start": "10:20", "end": "10:45", "track": [ "GraphQL" ], "room": [ "Doddington Forum 1" ], "title": "GraphQL as the Execution Layer for AI Agents", "speakers": [ { "name": "Michael Staib", "title": "Founder", "company": "ChilliCream" } ], "facets": [ "contract_quality", "agent_readiness" ], "agent_readiness_dimensions": [], "why": "GraphQL spec direction (abstract is a GraphQLConf recap).", "abstract": "GraphQLConf is just over and so many things happened! There's too much good stuff to keep to ourselves. In this talk, we'll piece together the fragments that matter most: the RFCs shaping the future of the spec, the proposals up for discussion, and the grants funding the work. Whether you missed the conference or just want a curated recap, you'll leave with a clear picture of where GraphQL is heading — and concrete ways to get involved, no matter how much time you have to give." }, { "id": 4110, "day": "2026-10-01", "type": "Talk", "start": "10:20", "end": "10:45", "track": [ "OpenAPI" ], "room": [ "Doddington Forum 2" ], "title": "OpenAPI Upgrades: the what, the why, and the how", "speakers": [ { "name": "Lorna Mitchell", "title": "API Architect", "company": "TM Forum" } ], "facets": [ "contract_quality", "contract_governance" ], "agent_readiness_dimensions": [], "why": "3.0 -> 3.1+: JSON Schema alignment, webhooks, $ref siblings.", "abstract": "Plenty of teams produced an OpenAPI 3.0 description years ago and have been using it ever since; the tooling has moved on, and they haven't. The upgrade to 3.1 and beyond is achievable and brings many benefits: proper JSON Schema alignment, nullable retired in favour of type arrays, webhooks described where they belong, and $ref that plays nicely with a description beside it. In this session you'll hear practical tactics for making the jump, what to expect along the way, and how to get the best from the tools that have appeared while you weren't looking. Find out what your API description could be doing for you instead of quietly rotting in a repository." }, { "id": 4123, "day": "2026-10-01", "type": "Talk", "start": "10:20", "end": "10:45", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "When the Agent Decides About a Person", "speakers": [ { "name": "Gulce Guleli ", "title": "HR Data and Insights Analyst", "company": "IG Group" } ], "facets": [ "accountability", "regulatory", "access_clarity" ], "agent_readiness_dimensions": [], "why": "An agent acting on a score about a named person; high-risk under the AI Act.", "abstract": "A retention model flags an employee as likely to resign. An agent reads that output and acts on it: a message to the line manager, or a change to who gets considered for a project. Nobody in that chain has spoken to the employee. Resignation prediction and candidate scoring are standard product features now, not experiments. Once a system produces a score for a named individual, the work stops being analytics and becomes a decision about that person. People data is the hardest governance problem in the enterprise. It is identifiable by definition, cohorts are small enough to reverse engineer, and employees cannot meaningfully opt out. Regulation has caught up: HR use cases sit in the high-risk tier of the EU AI Act, and GDPR Article 22 limits automated decisions that affect individuals. I have spent eight years in HR and people analytics across Big 4 consulting, high-growth tech and financial services. I'll cover what separates a system that keeps human judgement in the loop from one that only claims to. That includes the framework I use to make those calls, and a model we could not put into production because the world it learned no longer existed. This is not an HR talk. If you are building agents that act on data about people, the same constraints apply to you." }, { "id": 4000, "day": "2026-10-01", "type": "Talk", "start": "10:20", "end": "10:45", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "People-powered progress: OVO’s collaborative blueprint for responsible AI use", "speakers": [ { "name": "Mark Buss", "title": "Digital Sustainability Manager", "company": "OVO Energy" } ], "facets": [ "accountability", "outside" ], "agent_readiness_dimensions": [], "why": "Co-created responsible AI use at an energy retailer.", "abstract": "As a UK energy retailer with ambitious net-zero targets, OVO’s strength lies in our mission-driven team. When we introduced AI as a key tool to accelerate innovation as part of this year's targets, we knew we needed to start an open dialogue with our teams to ensure its use aligned with our shared values. This led to a journey where Tech & Sustainability teams partnered with people across our offices. Through collaborative workshops, we aim to transform initial concerns into a co-created strategy. This is how we’re harnessing technology and teamwork to build a more sustainable future." }, { "id": 4100, "day": "2026-10-01", "type": "Talk", "start": "10:45", "end": "11:10", "track": [ "GraphQL" ], "room": [ "Doddington Forum 1" ], "title": "Explaining GraphQL Federation to my non-technical spouse", "speakers": [ { "name": "David Stutt", "title": "Senior Founding Engineer", "company": "WunderGraph" } ], "facets": [ "contract_quality" ], "agent_readiness_dimensions": [], "why": "GraphQL federation explained. No abstract.", "abstract": "" }, { "id": 4111, "day": "2026-10-01", "type": "Talk", "start": "10:45", "end": "11:10", "track": [ "OpenAPI" ], "room": [ "Doddington Forum 2" ], "title": "The State of Proposals for API Security Profiles in OpenAPI", "speakers": [ { "name": "Chris Wood", "title": "Principal Architect", "company": "Ozone API" } ], "facets": [ "agent_readiness", "regulatory", "contract_quality" ], "agent_readiness_dimensions": [ "auth_clarity", "protected_resource_metadata", "auth_scheme_strength" ], "why": "Security profiles (FAPI, GNAP, OpenID Discovery) in OpenAPI.", "abstract": "OpenAPI Security Scheme Objects are a point-in-time snapshot of security that is increasingly duplicated in dedicated security profiles like FAPI and GNAP. OpenAPI needs a way to align more closely with security requirements, while also dovetailing with existing approaches like OpenID Discovery, which allows clients and agents alike to scaffold code and make runtime decisions meaningfully. This session walks through the state of the current proposals to enhance the OpenAPI Specification and unlock these capabilities. It will describe the proposed additions to OpenAPI, and how they might improve how security requirements are represented in OpenAPI in the future. We'll touch on the proposal's status and demonstrate meaningful examples showing the value of these improvements." }, { "id": 4016, "day": "2026-10-01", "type": "Talk", "start": "10:45", "end": "11:10", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "From API Traffic to Agent Execution: Why AI Agents Need a Different Gateway", "speakers": [ { "name": "Gbadebo Bello", "title": "Senior Developer Relations Advocate", "company": "Postman" } ], "facets": [ "agent_readiness", "operational_transparency" ], "agent_readiness_dimensions": [ "delegated_identity", "auth_clarity", "rate_limit_signal" ], "why": "From governing API traffic to governing agent execution.", "abstract": "AI agents do not behave like traditional API clients. They choose tools dynamically, call models and services in changing sequences, carry context across steps, and can retry or act autonomously. That makes the familiar request-by-request controls of an API gateway necessary—but no longer sufficient. This talk explores the shift from managing API traffic to governing agent execution. We’ll look at the identity, authorization, tool discovery, context, reliability, cost, and observability challenges that emerge when agents work across LLMs, MCP servers, APIs, and other agents. Then, through a live Banking investigation demo, we’ll compare a direct agent integration with a gateway-mediated path to show what changes when control is applied across the whole execution flow—not just at a single endpoint." }, { "id": 4091, "day": "2026-10-01", "type": "Talk", "start": "10:45", "end": "11:10", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "Beyond the Demo: Running AI Agents in Production to Modernize Enterprise APIs", "speakers": [ { "name": "Vedran Cindrić", "title": "Founder & CEO", "company": "Treblle" } ], "facets": [ "agent_readiness", "contract_quality", "operational_transparency" ], "agent_readiness_dimensions": [ "agent_skills", "mcp_server" ], "why": "Agents refactoring a legacy API estate in production.", "abstract": "This talk is built entirely on first-hand experience deploying AI Agents inside a large enterprise to refactor and modernize a legacy API landscape. Not in a sandbox, not in a lab, but in production, with real constraints, real data, and real consequences. I'll walk through what it actually takes: working within AWS Bedrock with sovereign data isolation requirements, spinning up sandboxed containers on AWS ECS in under 20 milliseconds, designing and optimizing agent Skills for code-level API refactoring, and integrating MCP servers for GitHub and Treblle to give agents the context they need to act intelligently. You'll leave with a clear picture of the architecture decisions that worked, the assumptions that didn't hold, and the patterns that made agentic modernization viable at enterprise scale." }, { "id": 4175, "day": "2026-10-01", "type": "Talk", "start": "10:45", "end": "11:10", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "From Principles to Standards: A Journey Toward Energy-Aware Streaming", "speakers": [ { "name": "Marisol Palmero", "title": "Technology & Standardization Initiatives Specialist", "company": "IEEE Europe Office" } ], "facets": [ "contract_governance", "outside" ], "agent_readiness_dimensions": [], "why": "How an energy measurement need becomes a standard.", "abstract": "What does it take to turn real-time energy measurements into actionable information across a complex value chain? Using video streaming as a use case, this session explores how an emerging technical need can enter the standardization landscape. Energy is only one part of the picture: measurements must be trustworthy, verifiable, secure and attributable across different stakeholders. With communities and SDOs looking at different pieces, the challenge is finding common ground through openness, collaboration and consensus—and exploring what can, and should, become a standard." }, { "id": 4172, "day": "2026-10-01", "type": "Talk", "start": "11:30", "end": "11:55", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "API Standards for AI Agents", "speakers": [ { "name": "Lorna Mitchell", "title": "API Architect", "company": "TM Forum" } ], "facets": [ "contract_governance", "agent_readiness", "regulatory" ], "agent_readiness_dimensions": [ "error_semantics" ], "why": "Layered standards: vocabulary, expected behaviour including errors, proven patterns.", "abstract": "AI agents are expected to act reliably across domains they've never seen, with inadequate documentation and inconsistent experiences. Layered API standards turn that from impossible into achievable: an agreed vocabulary, established expectations of behaviour including error cases, and patterns already proven in the real world. From the conventions that built the web to domain-specific specs, the standards stack is the solution. Come and see how your specifications become leverage for AI, not legacy overhead." }, { "id": 4101, "day": "2026-10-01", "type": "Talk", "start": "11:30", "end": "11:55", "track": [ "GraphQL" ], "room": [ "Doddington Forum 1" ], "title": "Teach Your AI Agent GraphQL", "speakers": [ { "name": "Dale Seo", "title": "Software Engineer", "company": "Apollo GraphQL" } ], "facets": [ "agent_readiness", "contract_quality" ], "agent_readiness_dimensions": [ "agent_skills" ], "why": "Teaching agents GraphQL. No abstract.", "abstract": "" }, { "id": 4023, "day": "2026-10-01", "type": "Talk", "start": "11:30", "end": "11:55", "track": [ "Apidays" ], "room": [ "Doddington Forum 2" ], "title": "Agent-Ready APIs: The Layer You're Not Building Yet", "speakers": [ { "name": "Sonali Goel", "title": "Software Development Engineer", "company": "Tesco Technology" } ], "facets": [ "agent_readiness", "developer_ergonomics" ], "agent_readiness_dimensions": [ "error_semantics", "idempotency", "event_surface_described" ], "why": "Agents do not read docs, do not handle ambiguous errors and do not wait on sync responses.", "abstract": "We've spent two decades designing APIs for humans. Clean endpoints, all built on a quiet assumption: that a developer, or an app a developer wrote, is on the other side. That assumption is broken. AI agents are the fastest-growing consumers of enterprise APIs and almost no API is ready for them. Agents don't read docs, don't handle ambiguous errors, and don't wait for synchronous responses. They need something fundamentally different from what we've been building for the past two decades. In this talk, I'll introduce the Agentic Action Layer — a practical middleware pattern that makes any existing API agent-ready without rewriting it from scratch. We'll cover the five things agents actually need from your API (that you're probably not providing), walk through of those components followed by an example wrapping a real API and unpack where emerging standards fits in." }, { "id": 4001, "day": "2026-10-01", "type": "Panel discussion", "start": "11:30", "end": "12:15", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "GreenOps Customer Showcase Panel", "speakers": [ { "name": "Kevin Leslie", "title": "Chief Revenue Officer", "company": "Greenpixie" }, { "name": "Mark Buss", "title": "Digital Sustainability Manager", "company": "OVO Energy" }, { "name": "Andri Johnston", "title": "Digital Sustainability Lead", "company": "Cambridge University Press & Assessment" }, { "name": "Tom Parry", "title": "Head of digital sustainability", "company": "Department for Environment, Food and Rural Affairs" } ], "facets": [ "access_clarity", "outside" ], "agent_readiness_dimensions": [], "why": "FinOps, tokenomics and sustainability customer cases.", "abstract": "How to Unlock the Double-Wins of AI and Cloud Sustainability. Using real-life case studies, this session explores how enterprise customers can deliver on the promise of business value and address IT Sustainability goals. How to support the demands of Responsible AI, FinOps and Tokenomics programs. An opportunity to learn directly from practitioners from both the public and private sector." }, { "id": 4127, "day": "2026-10-01", "type": "Talk", "start": "11:30", "end": "12:20", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "Securing APIs for the Agentic Era", "speakers": [ { "name": "Bharath Kumar", "title": "Founder & CEO", "company": "DigitalApiCraft" }, { "name": "Subbu Athikunte", "title": "DigitalAPI", "company": "Chief AI Engineer" } ], "facets": [ "agent_readiness", "regulatory" ], "agent_readiness_dimensions": [ "consent_identity", "dynamic_client_registration", "well_known_catalog" ], "why": "Discovery, consent rails and agent onboarding in banking, insurance and government.", "abstract": "Can your APIs trust an agent? Banking, insurance and government leaders on discovery, consent rails, agent onboarding and the controls needed before agents act." }, { "id": 4102, "day": "2026-10-01", "type": "Talk", "start": "11:55", "end": "12:20", "track": [ "GraphQL" ], "room": [ "Doddington Forum 1" ], "title": "Server Assisted Accessibility: Embedding UI Semantics Into Your GraphQL Schema + CI", "speakers": [ { "name": "Vanessa Johnson", "title": "Android Engineer", "company": "The New York Times" } ], "facets": [ "contract_quality", "contract_governance" ], "agent_readiness_dimensions": [], "why": "UI semantics in the GraphQL schema, enforced in CI. No abstract.", "abstract": "" }, { "id": 4132, "day": "2026-10-01", "type": "Talk", "start": "11:55", "end": "12:20", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "Composable Architecture for AI Systems: APIs, MCP, Skills, Sub-Agents, and Workflows", "speakers": [ { "name": "Erik Wilde", "title": "Head of Enterprise Strategy", "company": "Jentic & OpenAPI Initiative" } ], "facets": [ "agent_readiness", "contract_quality" ], "agent_readiness_dimensions": [ "agent_skills", "mcp_server", "spec_presence" ], "why": "APIs, MCP, skills, sub-agents and workflows: which mechanism belongs where.", "abstract": "AI is a powerful tool, but it is not the only tool, and most systems worth building are hybrids of AI and non-AI parts. That line between them doesn't sit neatly between domains. It runs inside every domain, and it should move as you learn. This talk looks at the mechanisms for connecting across that line: APIs, MCP, agent skills, and sub-agents, as well as at workflows as a way to move the line itself: when an agent keeps rediscovering the same business process, that process belongs in deterministic execution, not in a prompt. You'll leave knowing which mechanism belongs where, and how to keep shifting work out of AI as your system settles." }, { "id": 4113, "day": "2026-10-01", "type": "Talk", "start": "11:55", "end": "12:20", "track": [ "Apidays" ], "room": [ "Doddington Forum 2" ], "title": "AI Governance as Engineering: How Data Quality Determines Whether Your AI Can Be Trusted", "speakers": [ { "name": "Kingsley Okonkwo", "title": "Data Engineer", "company": "Ministry of Defense" } ], "facets": [ "contract_quality", "accountability" ], "agent_readiness_dimensions": [], "why": "Data quality decides whether AI can be trusted.", "abstract": "How the integrity of the data feeding AI systems determines whether those systems can actually be trusted to act." }, { "id": 4002, "day": "2026-10-01", "type": "Talk", "start": "12:15", "end": "12:40", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "You already have what you need: Open Source GreenOps in Practice", "speakers": [ { "name": "Julien Nioche", "title": "GreenOps Consultant and Open Source Veteran", "company": "DigitalPebble Ltd" }, { "name": "Diana Todea", "title": "Head of Developer Relations Engineering", "company": "VictoriaMetrics" } ], "facets": [ "open_source", "access_clarity", "outside" ], "agent_readiness_dimensions": [], "why": "SPRUCE: open-source GreenOps on Spark.", "abstract": "The environmental impact of cloud computing is surging, driven by AI workloads. While FinOps is now standard practice, its green counterpart—GreenOps—struggles due to a lack of transparent sustainability data from cloud providers. This talk introduces SPRUCE, an open-source platform built on Apache Spark to operationalise GreenOps at scale, enriching cloud usage data to quantify environmental impact and surface actionable insights. In the second part, Diana will explain how she built GreenOps dashboards from SPRUCE data using open source observability tools. She will also talk about related efforts in the ecosystem." }, { "id": 4103, "day": "2026-10-01", "type": "Talk", "start": "12:20", "end": "12:45", "track": [ "GraphQL" ], "room": [ "Doddington Forum 1" ], "title": "Houston, We Have a Partial Response: GraphQL Error Handling", "speakers": [ { "name": "Željko Kozina", "title": "Technical Architect", "company": "Chronomill" } ], "facets": [ "agent_readiness", "contract_quality" ], "agent_readiness_dimensions": [ "error_semantics" ], "why": "Partial responses and GraphQL error handling. No abstract.", "abstract": "" }, { "id": 4128, "day": "2026-10-01", "type": "Talk", "start": "12:20", "end": "12:45", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "Code, not tool calls: Deterministic, Reviewable AI Access to Enterprise APIs", "speakers": [ { "name": "Jens Neuse", "title": "CEO", "company": "WunderGraph" } ], "facets": [ "agent_readiness", "accountability", "contract_quality" ], "agent_readiness_dimensions": [ "operation_distinctiveness" ], "why": "Deterministic, reviewable, auditable code over tool calls.", "abstract": "The demo looked great, the audience was impressed. An agent was presented with a task, it called 3 different MCP tools and returned the result. Applause! 10x productivity, goal achieved! Except, the demo was actually inaccurate. The math was wrong, the agent aggregated the results incorrectly, and the final answer was wrong. However, it all sounded so confident that nobody questioned it. If important decisions depend on the results of a research, we cannot depend on \"\"hallucinating\"\" aggregations. This talk is about taking a different approach. The goal is to make AI access to enterprise APIs deterministic, reviewable, and auditable. How? The answer is to use LLMs for what they are best at: Writing code! We replace tool calls with code generation, inference with determinism, and a long context window with auditable artifacts." }, { "id": 4094, "day": "2026-10-01", "type": "Talk", "start": "12:20", "end": "12:45", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "AI - Spec-Driven Development: Lessons from £Billions in Payments", "speakers": [ { "name": "Cihan Dogan", "title": "Senior Software Engineer", "company": "Bally's Intralot" } ], "facets": [ "contract_quality", "contract_governance", "regulatory" ], "agent_readiness_dimensions": [], "why": "Spec first, then AI writes the code that moves money.", "abstract": "Letting AI write code is one thing. Letting it write code that moves real money is another. AI can write a week of code in an afternoon, but most teams still ship at the same speed they always did. Spec-driven development is what closed that gap for us: work that used to take months now takes weeks. In our payment team we changed the order of work. We agree the spec first, then let AI write the code. In this talk I walk through two real integrations shipped this way, a secondary Apple Pay and Google Pay provider built with GitHub Spec Kit, and a Bizum integration delivered in five weeks before the World Cup deadline, with the production numbers behind both. I will also cover what it takes to make this work at scale: keeping AI focused in a large codebase, where the process slows you down instead of speeding you up, and the honest pros and cons after a year of doing it in production." }, { "id": 4114, "day": "2026-10-01", "type": "Talk", "start": "12:20", "end": "12:45", "track": [ "Apidays" ], "room": [ "Doddington Forum 2" ], "title": "APIs as the Trust Layer for Global Carbon Markets: From Registries to Tokenised Assets", "speakers": [ { "name": "Oyediran Ojo", "title": "Senior Product Manager", "company": "S&P Global" } ], "facets": [ "contract_quality", "contract_governance", "discoverability", "outside" ], "agent_readiness_dimensions": [], "why": "APIs as the trust layer across carbon registries.", "abstract": "This talk explores how API‑driven infrastructure is reshaping the integrity and future of global carbon markets. Today’s voluntary carbon market is fragmented, opaque, and difficult to trust, with registries, standards bodies, project developers, and emerging tokenization platforms all operating in silos. As climate finance scales, this lack of interoperability becomes a systemic risk. The session argues that APIs are becoming the foundational trust layer that connects these actors, enabling verifiable data flows, preventing double counting, and supporting the transition from static carbon credits to programmable digital climate assets. Drawing on practical experience building carbon registry products, the talk shows how API‑first design can unlock transparency, automate MRV processes, and create the conditions for responsible tokenization. This matters because the credibility of climate action increasingly depends on the quality of the digital infrastructure beneath it. By treating APIs not just as technical interfaces but as governance tools, the industry can build markets that are transparent, auditable, and ready for the next decade of climate innovation." }, { "id": 4003, "day": "2026-10-01", "type": "Talk", "start": "12:40", "end": "13:05", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "Developing the world's first sustainability certification for cloud resources", "speakers": [ { "name": "Barton Finn", "title": "Project Manager - TCO Certified Cloud", "company": "TCO Development" }, { "name": "Emily Watson-Breeze", "title": "Purchaser Engagement Manager UK", "company": "TCO Development" } ], "facets": [ "access_clarity", "outside" ], "agent_readiness_dimensions": [], "why": "TCO Certified Cloud: a sustainability certification is a trust artifact.", "abstract": "As cloud adoption accelerates, organizations need practical ways to procure cloud resources more sustainably and to measure and report on their impacts. To address this need, TCO Development has developed TCO Certified Cloud, the world's first sustainability certification for cloud resources, with independently verified criteria covering climate, circularity, chemicals, supply chains and transparency. The first certifications are currently being piloted, with certified cloud resources expected in Q4 2026. The talk will share key insights from this journey and the future developments ahead." }, { "id": 4129, "day": "2026-10-01", "type": "Talk", "start": "12:45", "end": "13:10", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "DevOps-Driven API Governance in Practice: Open Source and Agent-Ready in Hours", "speakers": [ { "name": "Andrzej Jarzyna", "title": "API Evangelist", "company": "Independent Expert and author of \"RESTful APIs Design Patterns and Best Practices\"" }, { "name": "Krzysztof Madeński", "title": "API Consultant", "company": "PZU" } ], "facets": [ "contract_governance", "operational_transparency", "discoverability", "regulatory" ], "agent_readiness_dimensions": [], "why": "Spectral in CI, breaking-change detection, catalogue upkeep.", "abstract": "API governance is often discussed in terms of guidelines, standards, and central API catalogues. In many organisations, though, it is still too manual, too slow, or simply arbitrary. In this talk, we will show how to make API governance part of the DevOps process. Using tools like Spectral and CI/CD pipelines, we can automatically check whether APIs follow company standards, catch breaking changes early, keep APIs catalogued, and reduce drift between guidelines and what teams actually deliver. We will also show how to put in place a bare-bones, no-fuss governance setup in just a few hours, one that brings value quickly without becoming a big governance initiative. Finally, we will show how to treat AI agents as first-class citizens, so APIs are designed to be understandable and safe to use not only for developers, but also for agent-based consumers. We will present all of this in practice in a simulated environment." }, { "id": 4095, "day": "2026-10-01", "type": "Talk", "start": "12:45", "end": "13:10", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "A Valid API Call, An Invalid Outcome - Accountability in the Age of AI Agents", "speakers": [ { "name": "Afsha H", "title": "Software Engineer", "company": "Capgemini" } ], "facets": [ "accountability", "agent_readiness" ], "agent_readiness_dimensions": [ "reversibility_documented" ], "why": "A valid API call with an invalid human outcome.", "abstract": "AI agents no longer only recommend actions. They act through APIs. They can update records, provision infrastructure and trigger business workflows, potentially without a person reviewing each decision. APIs are becoming the hands through which AI agents affect people and organisations. Technical reliability matters, but it is no longer enough. A valid API call can create an invalid human outcome. Authentication, authorisation and schema validation can show that a request is permitted and correctly formed. They cannot tell us whether the action is wise, fair or ethical. This talk examines two responsibilities we cannot outsource to AI: ethical judgement and accountability, and human connection and transparency. What happens when a technically valid action produces an unacceptable outcome? When something goes wrong, can the agent understand what happened, challenge the decision and reach a human who can put it right? I will introduce SCOPE, a practical checklist for deciding what an agent should be allowed to do before it ever gets access. By the end, attendees will be well placed to design agentic API systems that are secure and reliable while remaining transparent, accountable and deserving of the autonomy we give them." }, { "id": 4173, "day": "2026-10-01", "type": "Talk", "start": "12:45", "end": "13:10", "track": [ "Apidays" ], "room": [ "Doddington Forum 2" ], "title": "Agent Control - How to take Agentic AI from Pilot to Production", "speakers": [ { "name": "Wynand Vorster", "title": "Pre-sales Manager", "company": "AppyThings" } ], "facets": [ "agent_readiness", "access_clarity", "regulatory" ], "agent_readiness_dimensions": [ "mcp_server", "operation_distinctiveness" ], "why": "API Products to stop tool sprawl; REST/SOAP -> MCP at MTN.", "abstract": "Moving agentic AI from pilot to production requires more than connecting an LLM to APIs. This session explores how an APIM platform can become an AI Gateway for governing models, securing backend tools, and exposing existing REST and SOAP APIs through Model Context Protocol (MCP). Using a real-world architecture deployed with MTN( Africa's largest Telco) , we’ll cover model routing and protection, MCP transcoding, and using API Products to prevent tool sprawl and reduce unnecessary context." }, { "id": 4104, "day": "2026-10-01", "type": "Talk", "start": "12:45", "end": "13:10", "track": [ "GraphQL" ], "room": [ "Doddington Forum 1" ], "title": "GraphQL with AWS AppSync - Building a Product API at HEMA", "speakers": [ { "name": "Ivan Jančić", "title": "Solutions Architect", "company": "Brighting" } ], "facets": [ "contract_quality" ], "agent_readiness_dimensions": [], "why": "GraphQL with AppSync at HEMA.", "abstract": "I will talk about \"pioneer phase\" of GraphQL adoption at HEMA and using AWS AppSync for the composable commerce migration, where I will share the raw and honest lessons learned when introducing a new tech stack to a 100-year-old Dutch icon." }, { "id": 4096, "day": "2026-10-01", "type": "Talk", "start": "14:10", "end": "14:35", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "From APIs to Agent(s): The New Control Plane for Digital Enterprises", "speakers": [ { "name": "Sivaraj Lenin", "title": "Senior Product Manager API Management", "company": "IBM" }, { "name": "Raghuram Tadipatri", "title": "Vice President - Hybrid Integration Software", "company": "IBM" } ], "facets": [ "contract_governance", "agent_readiness" ], "agent_readiness_dimensions": [], "why": "API management + integration + AI as one control plane.", "abstract": "APIs are no longer just endpoints in application architectures. They are becoming the decision and execution fabric for AI agents operating across enterprise systems. As organizations adopt agents, MCP-based tool access, and distributed AI systems, the challenge is no longer just exposing APIs—it’s managing how intelligent systems interact with them at scale. In this session explore the next stage of API evolution—where API management, integration, and AI converge into a unified control plane for the enterprise. We’ll learn emerging patterns such as federated gateways, lightweight runtimes, and AI interaction governance, and what they mean for architects, developers, and business leaders." }, { "id": 4131, "day": "2026-10-01", "type": "Talk", "start": "14:10", "end": "14:35", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "AI exposes 0.1x organisations", "speakers": [ { "name": "Joe Lewis", "title": "Vice President of Product Engineering", "company": "Envision Pharma Group" } ], "facets": [ "accountability" ], "agent_readiness_dimensions": [], "why": "Organisational change around engineers.", "abstract": "Why leaders must change the environment AROUND engineers, not just the Engineers themselves Lots of conference talks talk about how easy it is to put in place agentic AI, and have it write all your organisation's code. But this is harder in practice than in startups, but you can indeed change the culture of your engineering organisation, and move close to this. You just need the right plan, be aware of the challenges, and manage it like any change transformation Some options: 'The New Developer DNA: Embedding AI Literacy into Your Organisational Identity' or 'The Productivity Playbook: Raising the Floor and the Ceiling of Enterprise Engineering' or 'Culture Eats AI for Breakfast: Why Your Transformation Strategy Starts with People, Not Code'." }, { "id": 4115, "day": "2026-10-01", "type": "Talk", "start": "14:10", "end": "14:35", "track": [ "AsyncAPI" ], "room": [ "Doddington Forum 2" ], "title": "The Storyteller Agent: Beyond Code, Into Context", "speakers": [ { "name": "Eduardo Maldonado Fonseca Silva", "title": "Senior DevOps Engineer", "company": "" } ], "facets": [ "contract_quality", "discoverability", "agent_readiness" ], "agent_readiness_dimensions": [ "event_surface_described" ], "why": "AsyncAPI + ownership + capabilities as context for agents.", "abstract": "We're entering an agentic era where AI agents can generate code, review pull requests, and automate development tasks. But to become trusted collaborators, they need more than source code—they need context. In this talk, I'll explore how AsyncAPI can serve as the semantic foundation for building AI agents that understand an event-driven platform, not just its APIs. By combining AsyncAPI specifications with platform metadata, agents can reason about services, events, schemas, dependencies, ownership, and business capabilities as a connected story rather than isolated technical artifacts. Instead of answering only developer-focused questions such as \"\"Which service publishes this event?\"\", these agents can also explain why an event exists, how it supports a business process, which teams depend on it, and what the downstream impact of a change might be. This shared understanding helps bridge the gap between platform engineering and product teams, enabling more informed technical and business decisions. I'll present the architecture behind a Storyteller Agent, discuss the design choices involved in building a semantic knowledge layer on top of AsyncAPI, and illustrate how these concepts can be applied in practice using examples from an event-driven platform. Attendees will leave with practical ideas for transforming AsyncAPI from documentation into a knowledge layer that enables the next generation of intelligent platform agents." }, { "id": 4004, "day": "2026-10-01", "type": "Talk", "start": "14:10", "end": "14:35", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "Green Agentic AI: Optimizing Cost, Energy and Carbon", "speakers": [ { "name": "Navveen Balani", "title": "Executive Director", "company": "Green Software Foundation" } ], "facets": [ "access_clarity", "agent_readiness", "outside" ], "agent_readiness_dimensions": [], "why": "Cost, energy, carbon and water per agent task; retries and context as levers.", "abstract": "Agentic AI can consume orders of magnitude more compute than a simple chat interaction, making efficiency a design requirement, not an afterthought. This session introduces Green Agentic AI: how to measure cost, energy, carbon and water per task, then systematically drive them down. Using SCI for AI, emerging SEI and SWI metrics, and a six-stage agent lifecycle, the session shows the design levers across models, context, tools, retries, memory and scheduling, grounded in GSF standards and a practical Monday-morning checklist." }, { "id": 4105, "day": "2026-10-01", "type": "Talk", "start": "14:10", "end": "14:35", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "Architecting Sovereign AI Agents: Bridging Regional LLMs and Enterprise APIs on Hybrid Cloud", "speakers": [ { "name": "Daniel Oh", "title": "Senior Principal Developer Advocate", "company": "IBM" } ], "facets": [ "access_clarity", "agent_readiness", "open_source" ], "agent_readiness_dimensions": [], "why": "Sovereign agents, regional LLMs, hybrid cloud.", "abstract": "In 2026, the London tech landscape is defined by a push for data sovereignty and the rise of localized LLMs tailored for Southeast Asian languages and cultures. For platform engineers and architects, this creates a complex challenge: how do you orchestrate autonomous agents that are both culturally relevant and enterprise-ready? This session explores the intersection of Agentic AI, Open API ecosystems, and the hybrid cloud architectures necessary to keep data local and secure. Daniel Oh shares deep technical insights into building \"\"Sovereign Agents\"\" that use Open API standards to communicate across fragmented legacy systems and modern cloud services. We will dive into the role of Kubernetes and Knative in managing the lifecycle of AI agents, ensuring they remain cost-effective and scalable without vendor lock-in. You will leave this talk with a practical roadmap for deploying agentic workflows that leverage regional intelligence while adhering to the rigorous security and regulatory frameworks of the Singaporean digital economy. Discover how an API-first approach allows your organization to lead the next wave of AI innovation without sacrificing control or compliance." }, { "id": 4106, "day": "2026-10-01", "type": "Talk", "start": "14:35", "end": "15:00", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "Sovereign AI - Governing Enterprise AI Without Losing Control", "speakers": [ { "name": "Anuj Kumar", "title": "Principal Product Manager", "company": "Axway" } ], "facets": [ "regulatory", "access_clarity", "accountability" ], "agent_readiness_dimensions": [], "why": "Sovereign AI governance and regulatory requirements.", "abstract": "AI at scale demands more than innovation. Learn how to drive rapid adoption while protecting your data, meeting regulatory requirements, and building lasting trust." }, { "id": 4134, "day": "2026-10-01", "type": "Talk", "start": "14:35", "end": "15:00", "track": [ "Apidays" ], "room": [ "Hardwick Hub" ], "title": "If the developer portal is dead, what comes next?", "speakers": [ { "name": "Kristof Van Tomme", "title": "CEO, co-founder", "company": "PRONOVIX Developer Portals" } ], "facets": [ "developer_ergonomics", "discoverability", "agent_readiness" ], "agent_readiness_dimensions": [ "well_known_catalog" ], "why": "The developer portal's identity crisis.", "abstract": "Developer portals are in an identity crisis." }, { "id": 4116, "day": "2026-10-01", "type": "Talk", "start": "14:35", "end": "15:00", "track": [ "AsyncAPI" ], "room": [ "Doddington Forum 2" ], "title": "Designing Governance for Sustainable Open Standards", "speakers": [ { "name": "Thulisile Sibanda", "title": "Senior Community Manager", "company": "AsyncAPI Initiative" } ], "facets": [ "open_source", "accountability" ], "agent_readiness_dimensions": [], "why": "Five pillars of governance for open standards: transparency, shared ownership, succession, participation, accountability.", "abstract": "Successful open source standards require more than strong technical specifications. While specifications define how systems communicate, governance determines how communities collaborate, make decisions, and evolve. As standards mature, governance challenges can become a greater risk than technical ones: decision-making becomes concentrated among a few people, ownership becomes unclear, working groups lose momentum, and maintainers become overwhelmed. Left unchecked, these patterns can limit participation, slow innovation, and threaten long-term sustainability. This talk introduces the five pillars of sustainable governance: transparency, shared ownership, succession, participation, and accountability. Drawing on real-world examples and lessons from the AsyncAPI community, we’ll explore common governance anti-patterns and practical approaches for building resilient, community-driven projects. Whether you’re new to AsyncAPI, an existing contributor, a maintainer, or someone building another open-source standard community, you’ll learn practical ways to identify governance risks, strengthen contributor pathways, distribute responsibility, and create sustainable leadership models. The goal is not simply to define governance structures, but actionable strategies for identifying governance risks, strengthening contributor pathways, distributing responsibility, and creating leadership models that allow open standards communities to thrive beyond individual leaders." }, { "id": 3972, "day": "2026-10-01", "type": "Talk", "start": "14:35", "end": "15:00", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "Leading Sustainable AI from Strategy to Scale", "speakers": [ { "name": "Malini Jeyachandran", "title": "Lead Software Engineer", "company": "Lloyds Bank" } ], "facets": [ "accountability", "regulatory", "outside" ], "agent_readiness_dimensions": [], "why": "Sustainable agentic AI in UK banking.", "abstract": "Drawing on experience leading a pioneering customer-facing agentic AI proposition in UK banking, this session explores moving from strategy to production while scaling AI without scaling environmental impact. I will cover proportionate model selection, reducing computational waste, and embedding responsibility throughout the AI lifecycle. Attendees will learn to connect environmental responsibility with engineering efficiency, moving beyond governance checkpoints to continuous leadership commitments for sustainable, trusted AI." }, { "id": 4097, "day": "2026-10-01", "type": "Talk", "start": "14:35", "end": "15:00", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "APIs Under Agentic Attack: Closing the Speed Gap", "speakers": [ { "name": "Paul Fearns", "title": "API Security Systems Engineer", "company": "Akamai" } ], "facets": [ "operational_transparency", "agent_readiness", "discoverability" ], "agent_readiness_dimensions": [ "rate_limit_signal" ], "why": "Agents that map and chain APIs at attack speed.", "abstract": "Autonomous AI agents can map APIs, learn behavior, exploit weak controls, and chain functions at scale. Explore emerging API risks and defenses: discovery, governance, and real-time abuse detection." }, { "id": 4098, "day": "2026-10-01", "type": "Talk", "start": "15:00", "end": "15:25", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "If an API falls over in the forest, does it make a sound?", "speakers": [ { "name": "Yousaf Nabi", "title": "Developer Relations", "company": "SmartBear" } ], "facets": [ "developer_ergonomics", "operational_transparency", "contract_quality" ], "agent_readiness_dimensions": [], "why": "Who a change impacts and missing or wrong docs: contract testing territory.", "abstract": "APIs are the backbone of our modern day internet, but they are often misunderstood, miscommunicated and lead to friction by their very design. Decoupled and distributed systems allowed for decoupled and distributed teams, expected to converge at any given point, to develop a whole. The make up of which leads to friction when A) A developer wants to make a change to their API, but doesn’t understand who they will impact, and how B) An integrator who wants to utilise the API, they have no, or incorrect documentation, they need to expend time exploring the APIs surface area and boundaries rather than its capabilities and features C) The end-user, invisibly beholden to the connective tissue, a spider web of API calls between distribute and potentially disparate systems, suffers a degraded experience, when they stems fails to meet their primary goals At the centre of all of these stories, is the question of intent, and semantic understanding of the ways our systems communicate. These contracts bind our systems allowing us to leverage this friction, in the way of contract testing. Ensuring the systems behave as intended, by creating a system of record, of intent of consuming services, and validation of the providing services honouring of that contract. By allowing us to apply pressure to our design, it enables us to build better, more tolerant systems. The providing service developers, can surface conversations with relevant teams with pin-point precision, on a field by field basis, ensuring conversations happen at the right time, with the right people. This friction is surfaced by machines, and allows humans to decide on the path forward, be it a big-bang deployment where everyone is informed, to an expand and contract pattern whereby we gracefully migrate consumers onto new fields / schemas, endpoints or APIs, before removing functionality. With a clear view of our distributed estate and methods to ensure system compatibility, we can reduce the size of the mental model required to facilitate change in a system, harnessing friction at the right time, in the right way." }, { "id": 4174, "day": "2026-10-01", "type": "Talk", "start": "15:00", "end": "15:25", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "Reporting won’t save you: why GreenOps needs leadership accountability", "speakers": [ { "name": "Mark Butcher", "title": "Founder", "company": "Posetiv" } ], "facets": [ "accountability", "outside" ], "agent_readiness_dimensions": [], "why": "GreenOps as a leadership ownership problem, not a reporting problem.", "abstract": "Most organisations do not have a GreenOps knowledge problem. They have a leadership problem. When environmental impact sits outside real ownership of budgets, strategy, delivery and governance, teams get stuck measuring and reporting rather than changing anything meaningful. This talk explores why that gap matters, why it blocks the shift from reporting to optimisation to avoidance, and how to start building the accountability needed to fix it." }, { "id": 4107, "day": "2026-10-01", "type": "Talk", "start": "15:00", "end": "15:25", "track": [ "Apidays" ], "room": [ "Doddington Forum 1" ], "title": "Risk is Governed. Data Resilience is Architected.", "speakers": [ { "name": "Siham Boutayeb", "title": "Sr Principal Solutions Engineer", "company": "Odaseva" } ], "facets": [ "accountability", "operational_transparency", "regulatory" ], "agent_readiness_dimensions": [], "why": "Board-level risk vs recovery-tested systems.", "abstract": "A reflection on the alignment — or misalignment — between board-level risk governance and the architectural reality of data systems. Risk is discussed in boardrooms. Resilience is implemented in systems. But too often, the two operate in parallel universes. What does it mean to treat resilience not as a technical feature, but as a strategic discipline? How do we translate regulatory accountability into architectural patterns? Where does operational resilience truly live — in policy documents, or in recovery-tested systems? And how do we bridge the gap between governance language and data-layer implementation? This session explores how to structure resilience as a strategic capability — one that is designed, measured, tested, and owned." }, { "id": 4117, "day": "2026-10-01", "type": "Talk", "start": "15:00", "end": "15:25", "track": [ "AsyncAPI" ], "room": [ "Doddington Forum 2" ], "title": "Passports for Payloads: Standardising B2B Event Pattern with AsyncAPI", "speakers": [ { "name": "Aravind Viswanathan", "title": "Senior Technical Architect", "company": "Opencast" } ], "facets": [ "contract_quality", "agent_readiness" ], "agent_readiness_dimensions": [ "event_surface_described" ], "why": "B2B event exchange standardised with AsyncAPI.", "abstract": "Within modern enterprises, file-based batch processing has largely been replaced by Event-Driven Architecture (EDA), enabling systems to exchange information in near real time through event streams. However, this transformation has stopped at organisational boundaries. Business-to-business (B2B) asynchronous data exchange continues to rely on decades-old integration patterns, with SFTP-based file transfers, nightly batch jobs, and Managed File Transfer (MFT) platforms remaining the dominant mechanisms for sharing data between enterprises. Although these methods reliably facilitated bulk data movement in the past, they struggle to meet today's integration requirements. Their batch-oriented nature introduces inherent latency, delays the propagation of business events, and concentrates processing into resource-intensive compute bursts. While some enterprises have adopted pull-based bulk REST APIs, this approach introduces its own set of challenges, including constant polling overhead, complex pagination, and increased system load. The optimal approach to B2B data sharing involves extending Event-Driven Architecture beyond organisational boundaries. This is, admittedly, an unconventional proposition: most enterprises deliberately avoid exposing internal brokers externally, even behind a gateway, given the risks of blast radius, backpressure propagation, and operational coupling to an external party's failure modes. This talk argues that with the right mediation layer and contract-first governance enabled by AsyncAPI , EDA can be extended across trust boundaries without inheriting these risks. This Tech talk will provide a practical framework for tackling this problem with a focus on: • Transition Challenges : Why moving to EDA is critical for B2B integrations, along with the infrastructure, security and governance challenges that come with it? • Overcoming the Challenges : Practical approaches and strategies for addressing the infrastructure, security, governance, and operational challenges involved in transitioning to EDA for B2B integrations. • Reference Architecture : How to use Proxies and Protocol Mediation gateways to safely ingest external payloads (Kafka , WebSockets , AMQP) into internal brokers while protecting network perimeters? • AsyncAPI Governance : How AsyncAPI can serve as the \"passport\" (contract) for enforcing strict schema validation, version control, and channel boundaries?" }, { "id": 4118, "day": "2026-10-01", "type": "Talk", "start": "16:00", "end": "16:25", "track": [ "AsyncAPI" ], "room": [ "Doddington Forum 2" ], "title": "Discover. Authorize. Observe. Building Agent-Ready Event Streams", "speakers": [ { "name": "Dakshitha Ratnayake", "title": "Director - Developer Relations", "company": "WSO2" } ], "facets": [ "agent_readiness", "discoverability", "access_clarity" ], "agent_readiness_dimensions": [ "event_surface_described", "auth_clarity", "rate_limit_signal" ], "why": "Agents on event streams: discover channels, authorize, observe spend.", "abstract": "Picture an AI agent subscribed to a real-time event stream, calling a large language model on every event that arrives. Over a weekend, across hundreds of thousands of events, it quietly burns through hundreds of dollars. Nothing is technically wrong: every event is valid, the agent is authenticated, and no alert fires, because every existing check was designed to monitor infrastructure, not how an autonomous agent consumes a stream. As autonomous agents become primary API consumers, event streams are where this shift bites hardest. Developers build agent systems on raw event streams and they work. But in production, three things break. Agents can't discover what channels exist without hardcoded configurations. Any agent can subscribe to any topic because there's no channel-level access boundary. And there's no visibility into what each agent is consuming or what it costs when a high-throughput stream triggers LLM calls on every event. This talk proposes three questions that define an agent-ready event platform: Is it discoverable? Is access scoped? Is consumption observable? This talk uses a primary scenario of agents consuming a WebSocket stream backed by Kafka to show what AsyncAPI v3 provides: describable channels, schema contracts, and security documentation that your gateway layer can enforce. We walk through a real spec snippet. We cover the governance patterns that address the failure modes above, what exists in tooling today, and what still requires custom implementation. And we look at where AsyncAPI fits alongside MCP and OpenAPI in the emerging agent stack." }, { "id": 4141, "day": "2026-10-01", "type": "Keynote", "start": "16:00", "end": "16:25", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "Closing keynote: Notes from a Tech Futurist: Software Development five years from now", "speakers": [ { "name": "Mark Boyd", "title": "Director", "company": "Platformable" } ], "facets": [], "agent_readiness_dimensions": [], "why": "Closing keynote, no abstract.", "abstract": "" }, { "id": 4177, "day": "2026-10-01", "type": "Talk", "start": "16:00", "end": "16:25", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "Considering Your Personal Energy Flow; AI Sustainability Actions you can Take", "speakers": [ { "name": "Emma Albury", "title": "Digital Sustainability Lead", "company": "Government Digital Service" } ], "facets": [ "outside" ], "agent_readiness_dimensions": [], "why": "Personal and technological energy; GDS case studies.", "abstract": "In response to the conference theme “Follow the Energy” , GDS Digital Sustainability Lead Emma Albury calls for greater consideration of energy expenditure - both technological and personal energy. Cutting through the AI hype she asks “Why are we using AI? What is worthy of our energy?” and “How can we use our energy more sustainably?\". This is a call for action as a tactic to combat the climate and AI doom many of us face and instead, she offers hope using government case studies and practical tips to exemplify “good” AI use and purposeful use of energy." }, { "id": 4119, "day": "2026-10-01", "type": "Talk", "start": "16:25", "end": "16:50", "track": [ "AsyncAPI" ], "room": [ "Doddington Forum 2" ], "title": "From Generated Specs to Published Contracts: Our AsyncAPI Journey", "speakers": [ { "name": "Velmurugan Manickavelu", "title": "Senior Solutions Architect", "company": "Aristocrat" } ], "facets": [ "contract_quality", "contract_governance" ], "agent_readiness_dimensions": [], "why": "Generated AsyncAPI specs are not published contracts.", "abstract": "When we first started documenting domain events, we used OpenAPI because it was already part of our engineering practices. It worked well enough to get us started, but as our event-driven platform grew, we found ourselves forcing a request/response specification to describe asynchronous messaging. We adopted AsyncAPI to publish event contracts across our backend platform, where events are consumed by game teams, Data Science, QA, and other backend services. Using Springwolf, we automatically generate AsyncAPI specifications from our Spring Boot applications. However, we quickly discovered that generated specifications were not the same as published contracts. They exposed Java implementation details, framework-specific wrappers, and internal metadata instead of the domain language our consumers expected. In this session, I'll share how we transformed generated specifications into published contracts. We'll cover the customisations we made to produce consumer-friendly documentation, how we integrated AsyncAPI generation into our CI pipeline, and how every contract change is automatically surfaced through a pull request for architectural review before becoming the published contract. This governance process helps us evolve event contracts safely across independently deployed services with multiple downstream consumers. I'll also discuss the limitations we encountered. Producer discovery still relies on manual annotations, which means some producers might be missed if they are not properly annotated, and we intentionally focused on documenting the business contract rather than every messaging platform capability. This is not a deep dive into the AsyncAPI specification. It's a practical engineering story about introducing AsyncAPI into an existing platform, the trade-offs we made, the mistakes we learned from, and how generated specifications became trusted contracts shared across teams. Audience Takeaways Recognise when OpenAPI is no longer the right fit for event-driven APIs. Learn how to customise generated AsyncAPI specifications to communicate in domain language rather than implementation details. See a practical CI-based governance approach for reviewing contract changes before publication. Understand the difference between generated documentation and a published contract. Learn where automation helps—and where manual governance is still required." }, { "id": 4142, "day": "2026-10-01", "type": "Panel discussion", "start": "16:25", "end": "16:50", "track": [ "Apidays" ], "room": [ "Grand Hall 2" ], "title": "Closing Panel", "speakers": [], "facets": [], "agent_readiness_dimensions": [], "why": "Closing panel.", "abstract": "" }, { "id": 3976, "day": "2026-10-01", "type": "Talk", "start": "16:25", "end": "16:50", "track": [ "GreenIO" ], "room": [ "Grand Hall 1" ], "title": "Fractional Influence - everyday 'greening' as a contractor in tech", "speakers": [ { "name": "Fiona MacNeill", "title": "Director (UX | Research | Strategy)", "company": "Macknowlogist" } ], "facets": [ "outside" ], "agent_readiness_dimensions": [], "why": "Everyday greening as a contractor.", "abstract": "In a fragmented tech workplace how might an ethical contractor make a difference? As a technologist and researcher who transitioned into UX, I found myself in a pickle. No one was hiring researchers because LLMs made code cheap. So, I became a part-time product manager. As a PM, I realised that I had an opportunity to make a difference to digital sustainability with my small business clients. From little things like tagging tickets for ‘sustainability’, content formats, UI design, and token use. To strategic decisions such as hyperscalers, compute, and grant funding." } ]