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.
Building an AI agent that works in a demo is easy. Building one that safely runs for hours, survives failures, executes code, and can be trusted in production is a very different challenge.
In this hands-on workshop, attendees will build an autonomous agent using the OpenAI Agents SDK and Temporal while learning how sandboxing and durable execution enable safe, reliable autonomy.
Participants will start with a simple agent capable of completing coding and research tasks. They will then progressively evolve it into a production-ready system by introducing durable workflows, isolated execution environments, retries, human approval steps, and recovery from failures.
Along the way we’ll explore two complementary concepts of sandboxing:
- Temporal’s Workflow Sandbox, which guarantees deterministic workflow execution.
- Agent Sandboxes, which provide isolated environments where AI agents can safely execute code, manipulate files, and iterate without risking production infrastructure.
Rather than focusing on prompting, this workshop focuses on engineering: how to design systems that continue working when networks fail, processes crash, or tasks take hours instead of seconds.
Everyone leaves with a working application and a set of reusable patterns they can apply to real-world agentic systems.