Looking for 5 Claude power users to test something I built for long-chat handoffs
I’m looking for 5 Claude power users to help me sanity-check something I built.
This came from my own frustration.
I use Claude for long project chats, and the same thing kept happening: the thread would get valuable because it had all the context, decisions, failed attempts, files/errors, blockers, and next steps… then eventually it would get too long, too messy, or annoying to continue.
Starting fresh always felt painful because the new chat didn’t know the actual state of the work.
So I built AI Know It All.
Right now it starts with Claude.ai exports. The Chrome extension lets you export/backup Claude chats, and the web app turns a long export into a handoff doc + “Continue This Thread” prompt for a fresh AI session.
To be clear:
- It is Claude.ai only today
- It does not import ChatGPT yet
- It does not bypass Claude limits
- It is not magic memory
- It is meant to help with the restart/recap tax when a long thread becomes painful to continue
What I need now is real feedback from people who actually have messy Claude chats.
I’m looking for 5 people willing to test it with one real long Claude conversation and tell me whether the fresh session actually starts useful.
You can try it here:
https://aiknowitall.ai
Chrome extension:
https://chromewebstore.google.com/detail/aboidpipkbpflopfakglopgdmmaidknb?utm_source=item-share-cb
You get 3 free starter outputs, so you can test without paying.
And if it genuinely helps and you want Pro, I have a Founding 25 code: FOUNDER19. It makes Pro $19/mo forever instead of $29/mo for the first 25 people.
But the main ask is feedback, not payment.
I want to know:
- Did it save you recap time?
- Did it preserve the right decisions?
- Did it catch failed attempts / “don’t retry this” context?
- Did the fresh session understand what to do next?
- What felt confusing, wrong, or untrustworthy?
- Would you actually use this again?
I’m looking for blunt feedback, not politeness.
If you try it, reply here with what worked, what broke, or what felt confusing.
The handoff problem is real. Once a long chat turns into a pile of decisions, failed attempts and half-finished context, starting over feels a bit like amnesia. What I keep noticing is that the friction starts even before the export, when you’re trying to explain the next step clearly enough so the next chat doesn’t drift off. That’s the same kind of capture problem DictaFlow handles well, because it gets the messy first draft out of your head fast.
This is a real problem. Long AI chats become valuable exactly because they contain the messy project state: decisions, failed attempts, files, blockers, and next steps. Losing that context during a restart is not just annoying, it breaks momentum.
The strongest positioning here is not “AI memory.” It is project handoff infrastructure for AI work. That makes the product feel more serious and easier to trust, especially for Claude power users doing real work inside long threads.
The name is where I’d be careful. AI Know It All is memorable, but it sounds a bit gimmicky for a tool handling serious project context, failed attempts, and handoff accuracy. If the product expands into ChatGPT, team workflows, project continuity, or multi-AI handoffs, the name may start working against the trust layer.
Xevoa .com would fit this better as a clean AI workflow brand. It keeps the product on the user’s side: continuity, control, and smoother handoffs, without making it sound like a novelty AI tool.
I’d pressure-test that before more Chrome extension users, docs, and paid plans lock around the current name.
Tried setting it up and I think the biggest challenge here may actually be onboarding trust/clarity rather than the export itself.
I hit a “Frame with ID 0 is showing error page” state pretty quickly and immediately had a moment of:
The interesting thing is that products like this depend heavily on confidence because users are trusting the tool with large/context-heavy workflows they already care about.
So even small unclear states during setup can create hesitation very quickly.
I’d probably pay a lot of attention to:
This is super helpful, thank you.
I think I see what happened here, and this is on me more than you.
I didn’t make the starting flow clear enough. The extension is currently built around starting from Claude.ai / an actual Claude chat state. If it gets opened from the wrong browser state, or the Claude tab/frame is not available in the way the extension expects, it can throw that ugly “Frame with ID 0 is showing error page” type of message instead of telling you what to do next.
That’s bad onboarding on my part.
The product should have said something more human like: “Open Claude.ai first, go to the chat you want to export, then run the extension from there.” And if the tab is in a weird state, it should explain how to recover instead of showing a cryptic frame error.
I’m going to tighten that up because for this kind of product, trust matters immediately. If the first failure state is confusing, that’s a real problem.
Really appreciate you calling it out. If you’re willing to try again after I clean up the guidance/error state, I’d genuinely value the second pass. But either way, this was exactly the kind of blunt feedback I needed.