(and a callout)

I was initially exploring collaboration with another API provider whose work in artifact generation caught my eye. The exchange didn’t go the way I hoped - and rather than stew, I built. The analyze tool can now pipe your entire context to a new API we've created that generates:

  • Svg Graphics (charts, simple handouts, etc)
  • Powerpoint Presentations (no image support due to copyright concerns but it creates attractive templates very quickly)
  • PDF Files (via markdown)

You’ll notice that the backend pre-renders and provides computer vision–driven feedback on the state and structure of the documents. In a future update, this API will be able to “fully self-drive,” like the code tool does. The backend will use visual QA signals to automatically edit artifacts, improving quality and aligning results more closely with intent. For now, we’re keeping it in human-in-the-loop mode so users can see the kinds of results they’re getting and help us keep our prompt engineering sharp.

Stay tuned — the “Probes” update is coming next. Since collaboration didn’t work out, we’ve taken it as a challenge to meet (and exceed) the capabilities we admired from afar. If you're reading this, I hope the results speak for themselves.

It's the Red Queen's race now. Good luck.