Goldman Sachs peers pay premium rates for AI training help

The most expensive AI on Wall Street may not have been the software. It may have been the people teaching bankers how to use it.

Felipe Sinisterra and Dave Wang, both former bankers, were charging financial firms $25,000 a day to train staff on AI workflows, and their calendars were already booked for two months. Their pitch was practical, not theoretical: use Gemini to analyze founder pitch videos, or use a web application that combined behavioral-analysis methods with transcripts to flag possible red flags. The message for Goldman Sachs employees was hard to miss. The market was no longer paying up for access to models alone. It was paying for translation, for the ability to turn AI capability into something a banker could actually use before a client call, a committee meeting or a deadline.

That shift matters because it changes what gets rewarded inside a bank. For analysts and associates, the edge increasingly comes from writing better prompts, checking outputs, and tying AI-generated insights back to real client problems. For vice presidents and managing directors, the test is whether AI actually saves time, improves quality and cuts avoidable errors. If a team can run faster but still produces sloppy work, the technology becomes a cost center. If it trims rounds of revision and speeds up preparation, it becomes operating leverage.

The pricing also showed how AI training had become part of the competitive arms race. If a Goldman rival was willing to spend tens of thousands of dollars a day on education and workflow redesign, then internal training, certification and playbooks were no longer side projects. They were part of how firms protected margins and kept junior staff from becoming generic middlemen between clients and machines. In that environment, access to AI coaching could easily become another internal advantage gap, separating teams with better support from teams left to figure it out on their own.

Goldman had already been moving in that direction. In January 2025, the firm rolled out a generative AI assistant to bankers, traders and asset managers. By the time Goldman launched it firmwide in June 2025, Reuters reported that about 10,000 employees were already using the GS AI Assistant. Goldman also said in 2024 that it was rolling out its first generative AI tool for code generation to thousands of developers. Chief information officer Marco Argenti has argued that companies need to prepare “AI natives” to shape the future of work, a sign that fluency with the tools was becoming a management priority, not a technology hobby. For Goldman workers trying to move up, the new status marker may be less about saying the firm is experimenting with AI and more about proving that AI changed how the work actually got done.