Office Hours as AI Training Infrastructure
Office Hours
This points to Office Hours expanding from an expert call marketplace into a supplier of scarce human judgment for AI labs. The same system it uses to find, verify, and schedule niche professionals for research calls can also route bankers, lawyers, doctors, and other specialists into model eval and fine tuning work, where accuracy depends less on cheap clicks and more on whether the rater actually knows the field.
-
The product fit is operationally straightforward. Office Hours already solves search, matching, credentialing, scheduling friction, and expert incentives. In AI training, those same steps matter because labs need qualified people to review outputs, score reasoning, and write better examples in narrow domains.
-
The closest comparables show why this can be a real business line. Handshake turned its network into an AI data labeling business built around PhD students and professionals, while Prolific expanded from research participants into pre vetted specialist raters and AI task workflows.
-
This also changes where Office Hours competes. Instead of only facing GLG or AlphaSights for expert calls, it can increasingly overlap with AI data vendors like Mercor, Prolific, and Surge, where the wedge is trusted expert supply rather than generic annotation scale.
The next step is for expert networks to become labor infrastructure for reasoning models. As model builders push into finance, healthcare, law, and other regulated workflows, the winning platforms will be the ones that can deliver verified specialists on demand, capture their feedback in repeatable formats, and turn expert access into recurring AI infrastructure revenue.