Building Self-Improving Loops with Temporal and Oodle

AI Agents are like the hero from the movie Memento - they write sticky notes (memory) and forget everything. In this talk, we go over automation needed to build self-improving loops for AI agents. The talk covers Observing agent traces writing evals building a dataset bring your own agent for evaluation on the datasets Running an experiment Prompt management Human-in-the-loop approval workflows It uses Temporal and Oodle for the demo, but ideas are generic and can be adapted for your stack. #temporal #observability #oodle #aiagents #aievaluation