Handshake Shifts to Credentialed Expert Network

Handshake AI matters because it turned a cyclical recruiting network into an on demand expert labor marketplace for frontier model labs. After Handshake slowed to about $190M in 2024, the new business tapped its existing pool of PhDs, postdocs, and graduate students for math, law, and science evaluation work, pushing total annualized revenue to about $280M by August 2025, with roughly $80M coming from AI work.

• This is a product adjacency, not a random pivot. Handshake already had university relationships, verified academic identities, and a large supply base. That let it start matching experts to AI training tasks without building a contractor network from scratch, unlike newer labor marketplaces.
  1 sacra 2 sacra
• The market moved from cheap crowd annotation to expensive expert judgment. Recent AI training demand has centered on reasoning, red teaming, and domain specific evaluation, where a PhD in physics or a lawyer is more useful than a generic clickworker. That shift favored Handshake over older volume driven labeling models.
  2 sacra 4 sacra
• The revenue quality is different from the old job board business. Expert labeling revenue includes large contractor payouts, so reported topline behaves more like marketplace GMV and carries lower gross margins, around 25% to 40%, versus about 80% for Handshake's software heavy recruiting product.
  1 sacra 4 sacra

Going forward, Handshake is likely to become less of a campus recruiting company and more of a credentialed expert network with multiple monetization paths. AI training is the first large wedge. Data quality tooling, model evaluation, and other expert driven workflows can extend the same network far beyond hiring.