OpenAI has recruited more than 100 former investment bankers from top Wall Street firms to help train artificial intelligence models capable of automating the routine work traditionally done by junior analysts, according to documents reviewed by Bloomberg. The secretive initiative, code-named Project Mercury, pays contractors $150 per hour to build financial models and teach AI systems to perform repetitive banking tasks such as Excel modeling and PowerPoint pitchbook formatting, reports Finance Magnates.

Participants include former employees of JPMorgan Chase, Goldman Sachs, and Morgan Stanley, as well as professionals from Brookfield, Evercore, and KKR, alongside MBA students from Harvard and MIT. Contractors receive early access to OpenAI’s developing technologies and are required to submit one fully formatted, industry-standard financial model each week.

Banking Industry Braces for AI Disruption

The project targets junior investment banking analysts, who typically work 80–100 hours per week building detailed Excel models for M&A, restructuring, and IPO transactions while constantly revising presentations — the kind of work immortalized in Wall Street’s “pls fix” meme. Such automation could significantly reshape entry-level roles in finance: researchers at Stanford University have already recorded a 13% decline in employment among 22–25-year-olds in AI-exposed sectors since late 2022.

Industry expert Zhanna Brantover of DHR Global warns that removing manual tasks may hinder the professional development of junior bankers, noting: “Reading and analyzing documents is a process that people need to learn.” Major banks are already integrating AI — JPMorgan Chase recently announced tighter hiring policies amid a “tailwind of productivity” driven by artificial intelligence.