AnchorStrategy
Turn meeting notes into a strategy report — no hallucination
My tool (AnchorStrategy) checks every claim in its reports against the actual source data before it ships — no citation, no output. That's the whole reason I built it.
A report came back with a hard error: the citation checker flagged a claim as fabricated. Except it wasn't. The AI had correctly combined two separate findings into one sentence, which the system is explicitly designed to allow. But the checker code, when given two sources for one claim, was only comparing against whichever one scored as "most similar" — and silently dropping the other. The detail it needed to verify happened to live in the source it dropped, so it threw a false alarm on something totally legitimate.
Nobody had written a test for "what happens when a claim cites two sources at once." It just never came up until a real user's real input produced a claim that needed to.
Root-caused it, fixed it, added a regression test for exactly that shape of input, redeployed same day.
Nothing dramatic, but it's the clearest reminder yet that these edge cases don't show up from my own testing — only from real usage finding the combinations I didn't think to check.
My tool (AnchorStrategy) checks every claim in its reports against the actual source data before it ships — no citation, no output. That's the whole reason I built it.
A report came back with a hard error: the citation checker flagged a claim as fabricated. Except it wasn't. The AI had correctly combined two separate findings into one sentence, which the system is explicitly designed to allow. But the checker code, when given two sources for one claim, was only comparing against whichever one scored as "most similar" — and silently dropping the other. The detail it needed to verify happened to live in the source it dropped, so it threw a false alarm on something totally legitimate.
Nobody had written a test for "what happens when a claim cites two sources at once." It just never came up until a real user's real input produced a claim that needed to.
Root-caused it, fixed it, added a regression test for exactly that shape of input, redeployed same day.
Nothing dramatic, but it's the clearest reminder yet that these edge cases don't show up from my own testing — only from real usage finding the combinations I didn't think to check.
After a lot of small fixes driven by feedback from this community, AnchorStrategy is live on Product Hunt today.
Quick recap for anyone new: it turns meeting notes into a strategy report, but every claim has to trace back to the actual source sentence — no evidence, no number. A few of you helped me find real edge cases in that citation-checking logic over the past couple weeks, which genuinely made the product better.
Would appreciate any support or feedback here: https://www.producthunt.com/products/anchorstrategy
Quick build note for anyone following along.
When I started AnchorStrategy, I assumed the hard part would be getting the AI to produce good strategic analysis — SWOT, Porter's Five Forces, that kind of thing. It wasn't. Models are already pretty good at that.
The hard part was building the opposite: a system that's willing to stop and say "insufficient basis" instead of filling a gap with a plausible-sounding number. Every instinct in how these models are trained pushes toward completing the pattern, not admitting it doesn't have enough to go on. Getting the guardrails to actually catch that — reliably, not just in the demo — took longer than the entire rest of the analysis pipeline combined.
Small proof it's working: I posted about this on IH last week, and the comments that got the most traction weren't about the frameworks or the output format. They were specifically about that boundary — people recognizing that "confident but wrong" is a worse failure mode than "honest but incomplete." One conversation from that thread is now turning into an actual business discussion.
Still in private beta, still learning what people actually want to pay for. More updates as I figure that out.
Comment