We Trust AI Agents That Break Every Day
summary
Survey finds 80.8% daily use of agents but only 49.1% in production, with 85.5% saying they trust output while 41.1% see issues daily and 22% lose progress to crashes. A Temporal workflow demo moves a $2,500 payout using a local Quen 36B 9B model, records every step in durable execution, survives intentionally killed workers, preserves state, avoids rerunning LLM tool calls and therefore no re-spending of tokens, and resumes cleanly.
AI-written summary · about
chapters (12)
- 0:00Intro
- 0:34Finance runs on long-running workflows
- 2:05The State of Development survey
- 3:16Gap 1: Usage vs. production
- 4:33Gap 2: Trust vs. reality
- 7:18Gap 3: Prototype vs. enterprise
- 8:55Gap 4: Seeing vs. recovering
- 10:53Demo: a payout that survives a crash
- 13:06Durable execution explained
- 18:12What the demo solved
- 20:10Who runs this in production
- 20:50Wrap-up
