Jev Daily

Milind S says a cloud agent fixes bugs before the team reads them

A described workflow with no merge rate attached, posted alongside a cloud plan launch.

Milind S says users of Mausbot report bugs without leaving the chat, and that a cloud agent has usually opened the fix before the team has read the email.

Milind S
@milindlabs
X
our cloud agent has usually opened the fix before we've read the email.
Nov 8, 2026 · View on X

The loop he describes goes like this. A user sees something broken and tells the bot in chat. The bot asks a follow-up or two, then sends a bug report with all the context. That report lands in the team inbox, where a cloud agent picks it up. The agent runs 24/7, tests the build on a boat.dev machine, and attaches a screenshot from the virtual machine to the pull request itself. The team then reviews and merges.

The post doubles as a launch announcement. Milind S says that with this, they are launching their cloud plan.

What is not in the post

No numbers. There is no merge rate, no share of reports the agent actually resolved, no false positive rate on bug reports that turn out to be user error, and no cost per report or per pull request. There is no time figure either, beyond the claim that the fix usually lands before the email is read, which is a comparison against how fast a human checks an inbox rather than against anything measured.

That matters because the interesting failure modes in this shape of system are exactly the ones a screenshot in a pull request does not surface. A chat bot writing its own bug report decides what counts as context, and a user describing a broken screen is not a reproduction case. The agent testing its own build on a virtual machine is testing against its own understanding of what was wrong.

For people building on it, read this as a described workflow from someone launching a paid plan, not as a measured result. The screenshot in the pull request is the part worth stealing regardless of whether you trust the rest, because it gives a human reviewer something to check the agent's claim against before merging.

Know what Jev is good at before you bet on it

One email a day. Under three minutes. What builders found out about Jev yesterday.

Get the next one by email

Jev, read daily so you do not have to. The builds, the benchmarks, the criteria that worked and the cases where it lost, from the people shipping on TypeSafe AI's System One model.