Cooley Built Its Own AI System on ChatGPT to Speed Up IPO Work

Law firm Cooley didn’t just adopt ChatGPT — it co-built a proprietary offering on top of it. OpenAI detailed how Cooley’s capital-markets lawyers and legal engineers used ChatGPT Work and Enterprise to build “GO Public,” a system of purpose-built AI agents that combines client-specific information with agent-powered research and the firm’s own market judgment. The first application accelerates Form S-1 drafting, letting working groups and management teams reach the strategic questions sooner instead of spending early-stage time on drafting mechanics. Cooley says the underlying architecture extends beyond IPOs to other equity and debt capital-markets transactions.

The model — a firm-specific agent system co-developed directly with the model provider, rather than a generic chatbot layer — is becoming a recognizable pattern in professional services. Legal AI company Harvey has built the staffing side of that same pattern at much larger scale: roughly 180 “legal engineers,” all former practicing attorneys with eight to ten years of experience, embedded across every customer deployment, paid $220,000-$320,000 in on-target earnings — competitive with mid-level associate salaries. Harvey has scaled past 1,400 customers in 60 countries, running more than 25,000 custom agents inside customer environments.

Cooley and Harvey are solving different problems — one firm building its own tool with a model provider, one vendor staffing deployments with domain experts — but both point to the same conclusion for professional-services firms watching from the sidelines: document-intensive, high-stakes workflows like due diligence and regulatory filings are exactly where custom-built agent systems, not off-the-shelf chat interfaces, are proving out first.