Most teams building AI workflows reach for agents and skills, and end up with something probabilistic from end to end.

The agent calls whatever tool it decides to call, reaches for an expensive model on a step that needed a lookup, and runs as a black box: hard to validate, hard to predict, hard to cost out. This means a human has to check every output, or AI gets narrowed to low-risk work where a wrong answer costs nothing.

Join us and Barndoor AI for a live session on separating the one step that needs AI reasoning from the rest of the process, paired with an AI gateway that governs access, spend, and visibility across your environment.

What you'll take away:

  • How to evolve beyond skills. Why patching skills to chase reliability has a ceiling, and what comes next.
  • Inference only where it's needed. The steps that need judgment get AI. Everything downstream runs the same way, every time.
  • Full visibility. Debugging an agent workflow should look like debugging any other production system.
  • Scope, then distribute. Set the guardrails, access and tools a workflow needs, then govern how teams and agents run it.

Why you should attend:

  1. A process that worked yesterday might do something different today. Most workflows are probabilistic end to end, not just where judgment is needed.
  2. No consistent access control. Skills sit on one person's machine, and platforms that distribute a workflow govern it at the workflow level, not the model or MCP calls inside it.
  3. Cost and behaviour are invisible from outside the workflow. Nobody fully trusts it, a human keeps checking, and AI stays parked on work that does not matter much.

Your Speakers

Neil Mansilla VP of Platform, Barndoor AI

Neil runs platform strategy at Barndoor AI, managing MCP server integrations and customer rollouts. He previously led Developer Experience at Atlassian and worked as a software engineer in e-commerce, real estate and healthcare.

Jay Parisi VP of Solutions Engineering, Barndoor AI

Jay architects the solutions that help Barndoor customers govern and manage AI. He previously led enterprise solutions architecture at Twilio, managed solutions architecture at Coinbase, and worked in API management at MuleSoft.