AfterQuery's Reported $3.2 Billion Valuation Tests the Value of Expert AI Data

Mateega and Georgescu initially planned to build finance-focused AI agents. They say testing showed leading models failing on nuanced white-collar workflows, prompting a pivot toward capturing expert decisions and step-by-step reasoning rather than shipping another application on top of a model.

In its announcement, AfterQuery says nearly 100,000 verified practicing professionals contribute across engineering, medicine, law, finance and other fields. The company says it builds data-collection tools and workflows in-house, screens tasks so they are difficult but learnable, and runs post-training pipelines to measure benchmark changes before delivering datasets to customers. These are descriptions of its own process, not an independently reproduced evaluation.

The distinction matters because AfterQuery says it is producing demonstrations of how professionals complete tasks, not merely labels indicating whether an answer is correct. The company names Nvidia, legal-AI startup Legora and Korean AI lab Motif Technologies among projects or customers. Forbes says its data was used in Nvidia’s Nemotron models and that it worked with Thinking Machines Lab; neither the customer contracts nor spending levels are public.

Scarcity is the thesis—and the open question

The investment case rests on a familiar shift in AI economics. AfterQuery argues that web-scale text is increasingly exhausted, synthetic data cannot cover every edge case, and algorithms diffuse quickly. Domain-specific decisions from software engineers, lawyers and financial analysts could therefore become harder to reproduce than compute.

The comparison set shows why investors may assign infrastructure-scale values to this work. Forbes points to Scale AI’s 2021 rise and reports that Mercor was discussing a valuation of $20 billion with Nvidia. Those companies also show the limits of the analogy: human-in-the-loop suppliers compete for the same buyers, and a high valuation does not establish durable margins, exclusivity or customer retention.

AfterQuery’s founders say its software validates each contribution and that internal researchers demonstrate benchmark gains before a lab buys the data. That could give a buyer evidence about usefulness rather than a raw pool of labor. But the available material contains no benchmark methodology, baseline, retention rate, customer concentration, contributor compensation or pricing. It cannot show how much of the reported valuation reflects repeatable performance rather than scarce access to a fast-growing market.

What would settle the valuation claim

The next decisive evidence is a disclosed financing announcement or filing identifying the round’s terms, followed by audited revenue and margins. For the product thesis, customers or independent researchers would need to publish reproducible tests showing that AfterQuery’s post-training data improves real professional workflows, not only a selected benchmark. Until then, the $3.2 billion figure is a reported financing mark—and a signal about investor expectations—rather than proof that expert data is an enduring gatekeeper.