AfterQuery, a San Francisco company that pays doctors, lawyers and other credentialed professionals to build training data, has been valued at $3.2 billion. Gustaf Alströmer, a partner at Y Combinator, says no company in the accelerator's history has gone from launch to unicorn faster: the founders came out of the Winter 2025 batch, 18 months ago. Forbes reported the round first. AfterQuery's previous mark was $300 million, at its Series A. The company could not immediately be reached for comment.
The distance between those two numbers is the story. The only revenue figure AfterQuery has put in public is a $100 million annual run rate, disclosed at the same point as the $300 million valuation. Measured against it, the old price was three times revenue and the new one is thirty-two. Either the run rate has moved a long way since, or the multiple did all the work by itself.
What travelled with the news was a valuation, not a round. No size, no lead investor, no updated revenue number — which happen to be the three things that would settle which of those explanations is the right one.
AfterQuery belongs to the generation of companies built on the model Scale and Mercor established: recruit people with professional credentials — physicians, attorneys, specialists of every kind — and have them produce and check the data that labs train on. Its pitch goes a step past correct answers. The company says it trains models and agents to do the work the way practitioners do it, reproducing their approaches, their decisions and their reasoning. Its own phrase for this is "encoding the patterns, decisions and reasoning of the world's best practitioners."
The company has said it works with many of the largest labs. The customers it will name are Nvidia, Legora, and Motif Technologies, a Korean AI lab. The asymmetry is ordinary in this trade — a lab buying expert reasoning by the hour has more reason to stay quiet than the vendor selling it — but it does leave the disclosed customer list looking thinner than the price implies.
The "fastest unicorn" line deserves to be read as what it is: a Y Combinator partner grading a Y Combinator company against a record Y Combinator keeps. It is very likely true, and it is also a recruiting line for the next batch. The more interesting thing underneath it is the shape of the curve. An 18-month-old company's price moved more than ten-fold while its last public revenue number stood still, which means investors are paying for a belief about scarcity — that labs will keep needing human expertise they cannot generate for themselves — rather than for the income statement in front of them.
A short list of labs decides how much expert data gets bought, at what price, and for how long. AfterQuery is now priced as though all three answers are settled.