Hidden Gems in Public Sector Sales: Reading the Boardroom

If you sell into the public sector, the frustrating part is that everything you need to win is technically public — who the agency buys from, who approves it, who pushed back — but it’s scattered across thousands of board-meeting PDFs that no tool indexes and no rep has time to read.

The insight I built around: the useful signal isn’t the contract itself, it’s the discussion around it. Minutes record who introduced a deal, who voted against it, and sometimes why. That’s the account map a rep would normally spend a week assembling by hand.

So I built a pipeline that reads those documents and turns them into that map. For any agency, it’ll tell you who actually decides, which board members reliably block which kinds of deals, and what’s coming up for renewal — each answer tied back to the source page.

Live, I’ll query a corpus of 396 contracts I pulled from one Washington county. I’ll ask “who decides here, and what will they fight?” and it surfaces the director who introduced 83 of the last 396 contracts, and the board member who opposes anything that isn’t zero-emission — pulled straight from the minutes.

Two things I’ll show under the hood, because they’re the reason it works at all: one agent wrote every scraper that built this corpus, so adding a new agency is cheap instead of a custom job. And the system checks its own work without any labeled data — contracts have internal arithmetic that has to add up, so when an extraction is wrong, it catches and quarantines it. I’ll show that firing on real data.