For context: my team has 3 product engineers and a PM/product lead at an small series A startup.

Our PM has recently been asking to start contributing changes/features into our core (web) customer facing application using coding agents. The PM doesn't have any engineering background but has experience using Claude Code/Codex to develop hobby style apps in his free time for the past year or so.

I am curious of what other teams are doing and how its working out?

I like the fact that PMs are showing what they think for a product end-state rather than talking about it. One of our PMs showed a fully functional code for what we need to build. Problem? it is good on a sandbox, cannot be used for production as complexity comes from actual integration with many other systems which are kinda mocked in sandbox environment with sqlite, some json files etc.

Would I allow them to contribute to my app infrastructure repo? Honestly, I would not let anyone not fully understanding how it works to do so.

Even newly joined people to the team have developed similar habits of relying on AI to do "simple" things, like change an "exception" to "RuntimeException" that seems like a good thing to do but would actually cause problems as other parts of the code base relied on these "exceptions" for exception handling.

You could argue that if your tests are good enough you may not need to worry but that's BS, and this is my personal take.