Investor databases answer the question "who invests in my category?" That is the wrong question. Thousands of funds nominally invest in your category. The question that prices a round is: who is actively deploying, at my stage, at my check size, right now, and reachable through people who will vouch for me?
Stated thesis is marketing. Behavior is truth. A fund that says "seed to growth" but has written nothing but Series B checks for eighteen months is a Series B fund. A partner whose last four boards are all infrastructure deals has an infrastructure thesis, whatever the website says.
This gap between stated and actual thesis is not an edge case — it is the norm. Fund websites are updated once a year, if that. Partners rotate focus areas as markets shift. A fund can genuinely believe it still writes seed checks in your category while its actual capital, for the last six quarters, has gone somewhere else entirely. Nobody is lying; the website is just stale in a way a static list has no way to detect.
A fund that says "seed to growth" but has written nothing but Series B checks for eighteen months is a Series B fund.
The graph reads behavior instead: check cadence (how often, and how recently, a fund actually wires capital), stage drift (whether the stage of the checks is moving up or down over time), co-investor patterns (who a fund syndicates with, which says as much about its real risk appetite as any pitch deck), partner-level activity (attributing behavior to the individual, not just the fund brand), and the relationship edges that connect all of it back to you.
None of this is exotic data. Portfolio announcements, board changes, syndicate composition, and public filings are all observable — the work is in reading them as a continuously updated behavioral record instead of a one-time enrichment pass, and in re-weighting a fund's profile the moment its actual pattern diverges from its stated one.
Then the graph ranks — and because every input is evidence, every ranking is explainable. A 92% match is not a black-box number; it decomposes into the specific checks, the specific co-investors, the specific stage pattern that produced it, so the score can be argued with, not just trusted blindly.
We publish the reasoning with every match because explainability is not a compliance feature. It is what lets a founder walk into the room already knowing why the meeting makes sense to the other side of the table — which changes the conversation from "convince me you're relevant" to "confirm what the evidence already suggested."
