Tao Guo; Workshop
AI agents are increasingly expected to write code, execute commands, and solve complex problems autonomously. The challenge isn’t making them more capable—it’s making them safe and production ready.
In this hands-on workshop, you’ll build an AI agent using the OpenAI Agents SDK and Temporal, learning how sandboxed execution gives agents the freedom to act safely, while durable execution ensures work survives failures, restarts, human approvals, and long-running tasks.
You’ll leave with a working application and practical patterns for building resilient, production-ready agentic systems.
Jerome Brown; Workshop
Moving an AI agent from a fragile prompt playground to a production-ready cloud deployment requires a solid architectural foundation. Microsoft Foundry changes the paradigm by introducing a managed, containerised runtime for custom agent code via Hosted Agents, alongside centralised runtime tool discovery with Foundry Toolboxes.
This 3-hour, hands-on workshop is designed for engineers, architects, and technical leaders who want to learn how to stand up, secure, and scale production-grade agent environments. Leveraging pre-built, community-vetted labs, we will step through local verification, configure centralized toolboxes via the Model Context Protocol (MCP), and ground our agents using Foundry IQ retrieval. Best of all, we will focus heavily on architecture, configuration, and control-plane deployment—meaning no deep, from-scratch coding is required.
Otto Jongerius; Workshop
Most MCP setups in production have no audit trail, no policy enforcement, and no signed evidence of what the agent actually did. That's fine for prototypes — and it's exactly why security teams block these systems from going further.
This half-day workshop takes a working MCP setup and instruments it end-to-end: tamper-evident audit logs, Ed25519-signed receipts of every tool call, and a policy engine that blocks or allows at the proxy layer. You'll leave with a running pipeline you could demo to a security stakeholder on Monday.
We'll use agent-receipts (open-source, MIT) as the demo audit tool, but cover the alternatives — Pipelock, mcp-firewall — and discuss when each fits. The goal isn't to sell one tool; it's to give you a concrete, working starting point for the audit and policy conversation that's currently blocking your agentic features in production.
Caelan Huntress; Workshop
AI agents are changing the way work gets done.
As agentic AI systems become more capable, our real advantage will not come from technical proficiency. Our advantage will come from knowing how to manage AI agents well.
The ability to scope agentic projects, manage autonomous systems, evaluate their performance, and design the information environments that make agents effective - these are the skills of the future.
Josh Wulf; Workshop
Most codebases are not ready for LLM-assisted development. Pushing velocity before the substrate is ready produces regressions, drift, and loss of trust in the tool. Readiness is not about size or test coverage — it is whether the structural invariants are visible to the LLM or locked inside maintainers' heads. This workshop walks through Stage 1: working with an LLM to surface tribal knowledge, reclassify conventions as invariants, and migrate them through three levels of enforcement.