AI Pulse · Under the Radar

DeepSeek released its Harness agent framework, and within 48 hours four separate developers, none of them affiliated with DeepSeek, had already built tools on top of it. There’s a terminal interface with over 1,200 stars gaining almost 550 a day, a bilingual handbook with nearly 300 stars, a plugin index with almost 500, and a desktop wrapper with close to 250. Four creators, four different problems solved, same two-day window. You can see the first of these, the TUI repo here.

Here’s why that speed matters more than the tools themselves. When a new framework gets picked up this fast, it tells you the underlying thing is useful enough that people want to build on it before anyone tells them to. Nobody paid these developers. Nobody ran a marketing campaign. They saw a gap, in this case a better interface, documentation in two languages, a way to find plugins, a desktop option, and they filled it because the base framework was worth the effort.

For business owners, the lesson isn’t about DeepSeek specifically. It’s about what real traction looks like when a tool is genuinely good versus when it’s just well-promoted. If your team is evaluating new AI frameworks or considering open-sourcing something internal, this is the signal to watch. Stars per day, independent tooling, unpaid contributors showing up fast, that’s the tell. Announcements and press releases are not.

It also raises a question worth sitting with. If a tool inside your own stack got opened up to a wider audience, would it spark this kind of unpaid, fast-moving interest, or would it sit quietly? That’s part of why we think about this stuff when we design the way tools connect and get used, and it’s the kind of thing we build into an AI command centre, so useful pieces get surfaced and adopted instead of buried.