Crazy Agent Scale: Companies All-in on AI
Luke Wroblewski · 20 November 2026
Read the originalLuke Wroblewski's notes from an event on large-scale AI agent use show Sentry and Amp running most of their work through AI agents, with people reviewing what goes out.
Luke Wroblewski, a product designer, published his notes from Crazy Agent Scale, an event at the venture firm Sutter Hill Ventures. There, five builders showed how they run AI agents at large scale. David Cramer of Sentry, an error-monitoring company, and Quinn Slack of Amp, a company that makes a coding AI agent, described companies organised around AI agents.
- Sentry's AI agent, Junior, does nearly every job at the company, is its top contributor by commits, and credits the person who asked on every commit.
- Cramer said human attention is now the limit, so Junior sends its questions and pull requests ready for review into one inbox.
- Amp passed $100 million in annual recurring revenue with the same 21 people, each responsible for an area from deciding what to build to marketing it.
- An Amp agent groups about 9,300 bug reports from three months and drafts emails to affected users, and a person reviews every email before it is sent.
A product team can let an AI agent draft customer messages, provided a named person reads each one before it is sent. That is an inference and is not the speakers' claim. The page is a set of short notes from five talks, so each point is compressed.
Have a person review every customer email an AI agent drafts before it is sent.
Credit every commit an AI agent makes to both the agent and the person who asked for the work.
Derived by Working Surface from the article; more in the Playbook. Source line: Human attention is now the limiting factor.
21 November 2026: as AI agents take on investigation, design and drafting, the human work moves to choosing what the agents read and checking what they produce before it reaches customers.