But this is a problem you can fix with tools. Humans are inherently error-prone when it comes to reporting bugs, but computers aren’t. Imagine a tool that plugs into devtools and grabs: session recording + network logs + error traces + AI that writes a summary and repro steps.

A few years ago, we built a tool that captures all the information you’d get in a HAR file plus screen recording and packages it into a link for engineers. (Like a screen recorder plus DevTools). It’s called jam.dev and people use it to capture more than half a million bugs each month now (which we are really excited about!)

Today we added an AI inside that analyzes all of the session data and errors/traces in the console and network logs and writes a concise description of the bug and lists out the exact repro steps, complete with timestamps mapped to the screen recording. It’s built with Gemini 2.0 which we used while in experimental mode. We chose Gemini because of Gemini’s native ability to process videos (which other LLMs don’t have).

I’d love for you to try it out and let me know your thoughts, my email is [email protected].

Try it: jam.dev/ai Demo: https://drive.google.com/file/d/1g1RwoC8ITTTWR6BoIrgi_1YD6iD...

I think people are going to have a large role in testing software for a long time (as long as humans are the users, humans need to be testers too), but what doesn’t need to be manually done is collecting debug information and writing it up into a ticket. Like Kasparov and Deep Blue, the strongest collab is human + computer paired together –– humans to their strengths, computers to theirs.

By the way, we’re hiring engineers and working on building AI for debugging, de-duplication and auto-resolution and if these are problems that excite you, we’d love to chat: jam.dev/careers