💼 Why Big Foundation Models Won't Win Finance Without Specialized Infrastructure

👤 Gabe Stengel - CEO of Rogo (AI platform for data analysis in investment banking)
🎤 Patrick O'Shaughnessy - host of Invest Like the Best, founder of Positive Sum (venture fund)

📺 Invest Like the Best – 👥 121K subscribers
⏱ 1h 7m
🗓 22.09.26 (19h ago)
👁 18K

📝 Gabe Stengel, founder of the startup Rogo, explains how AI is transforming investment banking and private markets. They discuss the limitations of general-purpose models like ChatGPT, the critical role of infrastructure and data auditability, and emerging monetization models across fintech.

━━━━━━━━━━━━━━━

💡 KEY TAKEAWAYS

1️⃣ General AI labs won't build specialized infrastructure for finance
Anthropic and OpenAI are aiming for trillions in revenue, so chasing narrow vertical niches is not worthwhile for them. The winners will be products that build complex scaffolding: integrations with internal CRMs, secure data rooms, and regulatory compliance.

2️⃣ For investment analysis, auditability matters more than perfect accuracy
Financiers can tolerate a model's inaccuracies if all underlying assumptions and data sources are completely transparent. If an output cannot be verified and traced step by step for a regulator, the model is useless in production.

3️⃣ Private markets will become public-like through automated transactions
Getting a mortgage once required in-person visits to a branch, but today 40-50% of deals close online through services like Rocket Mortgage. Over the next 10 years, private capital will follow the same path: raising funds and selling businesses will come down to the click of a button.

4️⃣ Firm value will shift from human talent to systems and captured knowledge
Today, 90% of an investment bank's value resides in the minds of senior partners. In 10 years, 90% of capitalization will come from data, software, and cognitive systems that automatically replicate the expertise of top performers.

5️⃣ Within 18 months, every investment bank will make its core AI vendor choices
The applied AI market demands maximum aggression: companies must either monopolize their niche now or get consumed by larger players. The winners will absorb top talent and capital, creating a black hole effect.

6️⃣ Per-seat subscription models will give way to outcome-based pricing
Charging per token or per seat creates conflicting incentives with clients. The future of fintech software lies in capturing a cut of specific outcomes: a generated profitable investment idea or a successfully closed transaction.

💬 «100% of internal conversations in our company are recorded. If you want to turn individual productivity into enterprise value, you have to capture all context into a single company brain.»

🎯 WHAT TO DO WITH THIS
When building or adopting B2B software, focus less on the base foundation model and more on the integration layer connecting internal workflows and institutional memory.

💵 0.11$

@icereadspodcasts