When banks spent billions on artificial intelligence over the past two years, they assumed the hardware and software would unlock immediate productivity gains. Instead, a quieter crisis emerged: the people who bought the tools often have no idea how to use them. Enter Felipe Sinisterra and Dave Wang, two former investment bankers now charging up to $25,000 a day to teach senior financial staff exactly that.
Sinisterra and Wang are the founders of Wall Street Prompt, a boutique training firm that has become a go-to resource for some of the largest names in global finance. According to a detailed feature this week, clients include T. Rowe Price, Citigroup, and Bank of America, and the duo's calendar is fully booked for the next two months. Their service is straightforward: they show bankers how to apply generative AI models such as Anthropic's Claude, OpenAI's ChatGPT, and Google's Gemini to real financial tasks—tasks that the banks' own internal teams have not yet figured out.
The premium on practitioner credibility
The $25,000 daily rate is not arbitrary. It matches the approximate quarterly fee generation of a single managing director at a large investment bank, a subtle signal to procurement departments that the cost is too small to haggle over. It also surpasses the rates charged by big-four consulting firms for comparable training, reflecting the shift toward smaller, faster consultancies staffed by former practitioners who understand the nuances of financial workflows.
Sinisterra's background gives him instant credibility. He spent years at Goldman Sachs and Bank of America before leading fintech investments at SoftBank, where he deployed $2 billion and incubated several AI ventures. Wang's résumé is equally impressive: he was at Morgan Stanley and led crypto initiatives for SoftBank Latin America, and now sits on the advisory board of the Harvard Data Science Initiative. Their combined experience means they can speak the language of both technology and high finance—a combination that internal training departments often lack.
The gap between purchase and proficiency
Global banks have poured tens of billions into AI infrastructure, model licenses, and internal tooling on the explicit thesis that generative AI will reshape financial workflows. Yet the evidence from Wall Street Prompt's booking calendar suggests a painful truth: most organizations are still struggling with the granular, ground-level work of integrating probabilistic tools into a profession built on deterministic outputs. Earnings interpretation, market-analysis prompting, due-diligence synthesis, and pitch-deck review are all areas where analyst desks are operating at a fraction of what the underlying models can deliver.
In a demonstration described in the feature, Sinisterra and Wang showed senior bankers how to use Gemini's video-understanding mode to analyze a video pitch from a startup founder—a task that the bank's own staff had not attempted. Such live, novel use cases are the core of their offering. They bridge the gap between vendor documentation and practical application, showing bankers not just what the tools can do, but how to make them do it reliably in a regulated environment.
Why banks need outside help
The very structure of large financial institutions works against rapid AI adoption. Compliance departments demand verification of every output; risk teams need explainability; and senior executives, many of whom have spent decades relying on spreadsheets and intuition, are often skeptical of black-box models. Sinisterra and Wang's training sessions serve as a kind of cultural bridge, demonstrating that generative AI can augment judgment rather than replace it.
Moreover, the vendor landscape is evolving quickly. Anthropic, for example, has been actively pushing into financial services since early 2026, with partnerships including Moody's for data integration and full Microsoft 365 compatibility. As model vendors move closer to delivering plug-and-play financial workflows, the value of a bespoke prompting tutorial may decline. But for now, the gap between what banks have purchased and what they can use remains large enough to sustain a two-month waitlist at a premium price.
The long-term question
Whether the trainer category will compress is an open question. Sinisterra and Wang stay ahead by focusing on live, novel use cases that the vendor documentation does not yet cover. How long that gap stays open depends on how quickly model vendors integrate financial workflows directly into their products. For the moment, however, banks are paying for immediate competence, not future potential.
The irony is not lost on industry observers. The same institutions that spent billions on AI are now spending thousands more to learn how to turn it on. As one client testimonial noted, the actual training can be replicated by a moderately curious analyst with a corporate ChatGPT license and a weekend—but in the high-pressure environment of Wall Street, few have the time or permission to experiment. That is precisely the niche that Sinisterra and Wang have captured.