8+ Hour Livestream of AGNTCon + MCPCon EU Day 1 - Anthropic, OpenAI, AWS and More
Join us at AGNTCon + MCPCon NA https://hubs.la/Q04gbXLb0 Find all slides here: https://agntconmcpconeu26.sched.com/list/simple?iframe=no Timestamps 00:00:00 Pre-show and sponsor reels 00:13:00 Welcome to AgentCon and MCP Con Europe 00:20:00 Keynote: the five layer agentic stack and why open standards matter 00:32:00 Keynote: how far models have come and the MCP roadmap 00:40:00 Agentic messaging and skills over MCP 00:45:00 Keynote: reactive agents and what agents do while idle 00:59:00 Keynote: the production gap, kagent, and agentgateway 01:11:00 Morning keynotes wrap 01:13:00 Back room: what the keynote speakers cut from their talks 01:26:00 The view from the main stage 01:42:00 Talk: building a chief of staff agent at GitLab 02:03:00 Live demo: the chief of staff running the team 02:19:00 Talk: one bot and 16 agents in production for 5,000 developers 02:38:00 Debugging parallel agents and the silent failure 02:55:00 Talk: when not to build an agent 03:11:00 The router pattern and what replaced the chatbot 03:22:00 Talk: Zanzibar style permissions for agents with SpiceDB 03:40:00 Time bound grants and hierarchical access, live demo 03:51:00 Talk: AI agents that fix security findings in production 04:03:00 The process behind getting agent PRs merged 04:17:00 Back room interview: extracting meaning from backend data 04:23:00 Talk: agent identity, delegation, and scoped authority 04:35:00 Agent substrate, agent core, and Entra agent ID 04:45:00 Agent substrate demo 04:48:00 Back room: agentic coding, custom harnesses, and context 05:03:00 MCP server design and cutting token bloat 05:09:00 Agent memory, what to keep and what to forget 05:24:00 Tandem bike ride through Amsterdam with Dex 05:33:00 Interview: Block, Goose, and multiplayer agents in Slack 05:49:00 Beyond one product and one GUI 05:57:00 Afternoon keynotes begin 06:01:00 Keynote: what GitHub data says about agents writing code 06:08:00 New GitHub controls for agent proof of work 06:13:00 Keynote: training models for agentic work 06:22:00 Keynote: the AI control plane, MCP gateway, and sandboxes 06:33:00 Keynote: the agentic web and the end of the browser as we know it 06:40:00 MCP server, MCP app, and web MCP on one site 06:47:00 Back room interviews with GitHub and Google DeepMind 06:58:00 Talk: proving what your agent actually did 07:09:00 Mapping obligations, the EU AI Act, and tamper evident records 07:24:00 Showcase floor: AI Booster, Postman Fabric Gateway, and docs for agents 07:34:00 Talk: 90 days to agentic engineering 07:48:00 Skills, review, and rolling agents out across a company 07:56:00 What worked, what failed, and what to do earlier 08:00:00 Showcase floor: Traffic Labs 08:07:00 Back room: A2A, agent specialization, and observability 08:22:00 Being opinionated in a fast moving ecosystem 08:29:00 AgentCon Japan 08:36:00 Back room: agent identity, guardrails, and approvals 08:47:00 Talk replays and closing