Agent Baseline Demo Night

​We've watched agents escape sandboxes, drop databases, and trigger unauthorized production deploys. Agent behavior is unpredictable and teams that want to operate agents in production need to be able to confidently answer three key questions:

• ​What is operating, with what capabilities?

• ​Is it staying inside approved boundaries?

• ​Can you prove what happened and stop it?

​Agent Baseline, reference architecture for deploying and operating agents in production, answers those by outcome instead of by product category: Discover, Constrain, Authorize, Observe, Validate, Respond. It was first released by Docker, Keycard, and Snyk in August, and is vendor-neutral and open source.

​On the Tuesday night of AI Engineer Financial Services, we'll show you how you can use Agent Baseline as a guide to create production ready agents. We'll have a short panel to explore Agent Baseline together followed by six lightning demos, one per outcome described in Agent Baseline, to show you how to achieve these outcomes in practice.

​On the panel:

• ​Dick Hardt, Founder of AAuth and co-authored of OAuth 2.0 and JWT

• ​Ian Livingstone, Co-Founder and CEO at Keycard.

• ​others to be announced soon

​Moderated by Allie Howe, host of the Insecure Agents Podcast.

​What you'll see:

• ​Six live demos, one per outcome:

  • ​How to discover what's already running

  • ​How to constrain an agent's execution environment

  • ​How to authorize agent actions at runtime and grant access scoped to the task at hand

  • ​How to observe agent behavior and drift

  • ​How to validate agentic components

  • ​How to respond to a rogue agent

​Doors open at 5:30. Come if you're shipping agents into production and want to feel confident in what you've built.