How Confluent's playbooks run themselves.
Confluent's AI agents run in production. Not on someone's laptop. Confluent's Applied AI leaders on what happens to the builds that prove themselves, and what it takes to run them autonomously.
Save your seat
A recording is sent to everyone who registers.
What you'll walk away with
The last 10% is where the work actually is. Standing up an agent is fast now. The business decisions, the edge cases and the checks around them are what take the time.
Where a personal build has to stop. Every Confluent user has Claude Code, and personal agents are encouraged. Erik and Mike explain what has to be true before one goes near the business.
Why each new agent is quicker to build than the last. Every build leaves tools and skills behind. The next agent starts with parts already on the shelf.
How they decide what not to build. A lot of what lands in the queue is not an agent problem at all. Saying so early is part of the job.
Why a first agent is deliberately small. Low volume and a limited blast radius, on purpose. What they are really building is the builder.
A look under the hood. A live walkthrough of a pre-built Relevance multi-agent system: the agents, the processes, the data flows, the visual canvas. Then open Q&A.
Unlock the agentic ROI you promised your board
Relevance's platform maps the path from assisted AI to full autonomy. Real business impact is driven in L3/L4.
Delegate busywork like research and drafting to a copilot like Cowork.
Teach your copilot your playbooks. Powerful, but hard to govern.
Proven playbooks become governed agents that run autonomously.
Your agents run evals and swap models themselves. You lead strategy.
