AI Pulse · Under the Radar

A new tool for extracting and standardising AI coding chat histories has drawn attention quickly. In its first two days, it reached 321 GitHub stars, running at roughly 157 stars a day. The project pulls conversation data from Claude Code, Cursor, Windsurf, Aider, and Cline/Roo Code into a more consistent format. You can see the early project activity in the original repository.

Why should an owner care? Teams are doing more work inside AI assistants, especially technical teams. The useful context, decisions, prompts, errors, fixes, and working methods often sit inside separate chat logs owned by different tools. That makes it hard to understand how work is actually getting done, where time is being lost, or what good practice looks like across the business.

This is part of the session-mining trend already on our radar. The immediate takeaway is not that this particular tool is proven. It is only two days old. The signal is the demand behind it. Businesses want a clearer record of how people use AI at work, not just a list of licences they pay for. Once those sessions can be normalised, they can be reviewed for recurring problems, training opportunities, reusable workflows, and risk. This is the kind of thing we build into an AI command centre, bringing scattered AI activity into a view leaders can actually use.