Ask HN: Built an algorithmic forensic accounting tool
Over the last two months, I’ve posted a few times about a tool I built to automate forensic accounting for high-conflict divorces. (I recently wrote about using Python to automate a $5k forensic CPA job).
The tech is solid. It uses a spatial-grid OCR pipeline to ingest blurry bank PDFs, applies the court-mandated Lowest Intermediate Balance Rule (LIBR) using graph theory, and outputs a cryptographically sealed (SHA-256) dossier tracing commingled assets. Because of attorney-client privilege, it runs air-gapped in a Docker container (Sovereign Mode).
It replaces a 3-week, $15,000 CPA engagement with a 3-minute, $497 algorithmic trace.
But I am hitting a wall on the go-to-market side.
Lawyers are incredibly risk-averse. I am trying to get just one family law firm or forensic accountant to run a pilot completely free, off the record, run in parallel with their manual process to prove that the algorithmic output beats the CPA's output.
My questions for those who have sold into LegalTech or highly risk-averse B2B:
-Short of cold-emailing managing partners, how did you find your first early adopter?
-Should I be targeting family lawyers (who want leverage to win cases), or forensic accounting firms (who might see this as eating their billable hours)?
-How do you bridge the trust gap when you are an engineer, not a lawyer/CPA?
Also: If anyone here happens to be a family lawyer, forensic CPA, or is dealing with a complex divorce and wants to run a gnarly ledger through the system as a free pilot, my email is in my profile.