OpenAI has brought together over 100 former investment bankers to help train its AI to take on the routine tasks usually done by junior bankers.
This project, called Project Mercury, aims to automate financial modelling and other entry-level duties that are time-consuming for junior staff. The group includes former employees from major financial firms like JPMorgan Chase, Morgan Stanley, and Goldman Sachs.
These experts are paid $150 per hour to build financial models for transactions like IPOs and restructurings while also helping create prompts for the AI.
In return, they get early access to OpenAI’s tools, which aim to replace much of the grunt work done by junior bankers
What is Project Mercury?
The project shows OpenAI’s drive to make its AI models useful across many industries, including finance, consulting, legal, and technology. Even with a $500 billion valuation, OpenAI is still working toward finding a profitable business model.
An OpenAI spokesperson said the company works with various experts “to improve and evaluate the capability of our models across different domains,” adding that these experts are hired and paid through third-party suppliers.
Junior bankers or investment banking analysts often work over 80 hours a week, building financial models in Excel for mergers, buyouts, and more.
They also spend much of their time tweaking PowerPoint presentations, leading to the rise of Wall Street’s “pls fix” meme. As AI continues to advance, many startups are aiming to automate these tasks, which is making junior bankers concerned about job security.
Hiring process for ‘Project Mercury’
Applying to Project Mercury involves a nearly automated process. First, applicants have a 20-minute interview with an AI chatbot, followed by tests on their knowledge of financial statements and financial modelling.
Contractors working on the project are expected to produce one model per week, following detailed instructions. After submitting their work, they receive feedback and must fix any issues before their models are integrated into OpenAI’s systems.
Participants in the project come from a range of financial firms like Brookfield, Mubadala Investment, Evercore, and KKR, along with MBA students from top schools like Harvard and MIT.
Along with modelling, participants must ensure their work follows industry standards, such as proper formatting and presentation of percentages, like italicising numbers and adjusting margin sizes.