Each week the newsletter takes one question teams are working through, such as who approves what an agent built, and shows how real teams are answering it: the six articles that say most, with the line worth quoting from each.
You get one email to confirm, then one every Monday. Every email has a link to unsubscribe.
Your address goes to Resend, the service that sends the email, once you confirm it. It is used for the weekly and nothing else. Privacy

Weekly
We wanted voice AI in our payments app. Here’s what we cut.
Glomo's designers cut a voice assistant down until speech could never approve a payment, and four more pieces from this week draw the same kind of line: teams decide first what AI agents will not do, and keep that work for a person.
6 links · Agent ownership

Weekly
Judgement > Output
Cate Huston argues that AI has made output cheap, so people are now judged by the judgement behind their work, and five more pieces from this week agree: the scarce work is choosing what to build and checking what was built.
6 links · Changing roles

Weekly
How we review plans, not pull requests
Giant Swarm moved human review to the plan, and five more pieces from this week do the same: teams write down the plan, the question and the assumptions before AI agents build, where someone else can dispute them.
6 links · Decision records

Weekly
Who keeps what the agent reads?
Six pieces from this week on the written rules AI agents work from: who writes them, where the agent finds them and who keeps them true.
6 links · Agent context files

Weekly
What is an approval worth?
Six pieces from this week on human approval: when a person's sign-off on an AI agent's work protects anyone, and when it is only a button.
6 links · Human approval

Weekly
Who owns what the agent made?
Six pieces from this week on work made by AI agents: a named person still owns it, reviews it and answers for it.
6 links · Agent ownership

Weekly
Checked against what?
Six pieces from this week on checking work made with AI: what the reviewer compares it with decides what the review is worth.
6 links · Ground truth for review
Hello, I'm Wevy. Ask how product, design and engineering teams work with AI agents. The answer comes only from the 315 links on this site.
Claude picks the links and writes the answer from them only. Answers can be wrong. Do not type personal information: questions go to Anthropic. How Ask works · Privacy