Flowacts: I built a canvas to trace AI answers back to their sources

I built Flowacts because I kept running into the same problem when using AI for research and planning.

The answer would look useful, but after a while I could not tell what was actually based on the material I gave it. A note I wrote, a webpage I pasted, a PDF I uploaded, and something the model generated on its own could all end up inside the same response and look equally convincing.

That made the answer harder to trust.

I do not think there is a magic way to make AI stop hallucinating completely. But I do think the work should be easier to inspect.

So I built Flowacts, an AI canvas where sources, notes, and answers stay connected. You can start with a question, file, webpage, or note, then branch into new questions, maps, plans, comparisons, or decisions without losing where the ideas came from.

The part I care about most is not just getting another polished AI answer. It is being able to see what the answer is actually based on.

If an idea came from a webpage, PDF, or note, you should be able to trace it back. If something is just part of the model's reasoning or generated text, that should be easier to notice too. Not as a perfect truth machine, but as a way to keep the source, context, and thinking process visible while you work.

I built this independently, and now I’m sharing it with the Indie Hackers community.

When you use AI for research or planning, do you check where the answer came from before trusting it?

And if you do, what do you actually look for?