On the afternoon of Mar 3, 2026, Bolt.new’s build agent stopped responding for a few hours. Deploys queued up and never finished. Nothing told the people shipping on Bolt.new that anything was wrong. Their work just stopped moving, quietly.
The people who noticed first were the ones whose own users started asking why the app wasn’t updating. That is the exact thing that shouldn’t happen: your customers shouldn’t be the first to know.
What happened
Bolt.new’s status page reported degraded performance on its agent. Builds that normally finish in seconds sat in the queue. There was no error on the screen, no email, no alert. Just a quiet stall. For a few hours, anyone relying on Bolt.new to ship a change was stuck without knowing why.
What a watcher would have caught
A watcher checking Bolt.new’s public status every minute would have seen the agent flip from working to having problems within a minute of the upstream report, and said so in plain language, long before a customer noticed the app had stopped updating. That is the whole point of watching the tools you build on: you find out first, and you find out in words you can act on.
The takeaway
Every tool you build on can have a quiet afternoon like this. The fix isn’t to stop using Bolt.new. It’s to make sure that when it has a bad few hours, you hear about it from a calm plain-language heads-up, not from the person paying you.