Agents: Building Production AI Agents with Temporal
summary
Retool turns its Temporal-based workflow engine into a production AI agent system that runs LLM calls and tool executions as Temporal activities with workflows for state transitions, uses dynamic tool routing and step-level observability, and implements human checks via Temporal signals. A V3 planner groups nodes to cut 30 activities to six, yielding ~5x fewer activities, ~8x faster runs, saving about $9M/year across ~10M daily runs and reclaiming 735 years of cumulative latency.
AI-written summary · about
chapters (8)
- 0:00Introduction
- 1:07What is an AI agent
- 3:55Retool and Temporal
- 7:04Workflow architecture
- 13:46Building agents with workflows
- 16:54Human-in-the-loop workflows
- 20:43Future of AI agents
- 23:51Retool Assist and agents
related talks
24mDurable Agents: long running AI workflows in a flakey world
33mFrom Prototype to Production: Making Agentic AI Systems Reliable
57mWhen AI Agent Teams Work (And When They Just Create More Work)
08mReal AI Agents Need Planning, Not Just Prompting
34mAI Agents: Transforming Autonomous Decision-Making
42mUsing Temporal Workflows for Seamless Dev + Research Collaboration & Continuous Algo D
16mHow agents will unlock the $500B promise of AI
22mHOW TO BUILD YOUR OWN AI AGENT: THE COMPLETE ARCHITECTURE FOR ENGINEERING TEAMS