Goldman Sachs investment banking co-head Kim Posnett says artificial intelligence has lifted the baseline for client conversations, forcing senior bankers to deliver sharper advice because counterparts now arrive better informed. The shift is reshaping how the firm's most experienced dealmakers prepare and how its newest analysts learn the trade.
The meeting has changed
Posnett told me that chief executives, founders and investors now walk into rooms having already digested vast amounts of data. That means the value of a Goldman banker is no longer in assembling the facts but in interpreting them. She uses AI to compress research that once took hours into minutes, freeing time for the strategic judgment clients actually pay for. The dynamic applies across the table: when everyone is smarter going in, the conversation moves faster and the expectations climb.
Apprenticeship accelerated
The firm's $6 billion technology budget is not just a senior-banker tool. Junior analysts are sitting in on high-stakes meetings earlier because AI handles the grunt work, summarizing earnings calls, mapping industry trends, building first-draft models, that used to consume their first years. Posnett argues this compresses the apprenticeship curve: the youngest staff spend less time formatting slides and more time watching how seasoned advisers steer complex decisions. The culture that defines Goldman, she says, is learning by proximity, and technology is making that proximity happen sooner.
The balance sheet for skills
David Solomon's intern letter this year urged experimentation with AI. Posnett echoes that but adds a guardrail: master the tools, then invest heavily in the things they cannot do well. Ask better questions. Develop judgment. Communicate. Build relationships. Earn trust. She recently laid out five priorities for the incoming analyst and associate class: nail the technical foundations, sharpen the questions you ask, redeem the time AI gives back, lean into the accelerated apprenticeship, and double down on the human qualities that close deals. The logic is circular: you need the technical base to know whether the machine's output is right, and you only get that base through repetition and experience.
No substitute for the basics
Slide decks and financial models are increasingly automated. Posnett insists that does not make the underlying craft optional. If an analyst cannot explain the drivers behind a valuation, the fact that a model produced the number in seconds is irrelevant. The firm's bet is that AI raises the floor for everyone, but the ceiling still belongs to people who understand the business, the client and the market well enough to know what to ask next.