Our Conversation with OpenAI's Former Head of Sales: Enterprise AI Adoption and the Opportunity in Vertical AI
Key insights from our SF Tech Week event on why financial institutions are ready to buy AI now, how OpenAI scaled from $10M to $10B in revenue, and advice for founders building in vertical AI
We hosted the most popular fintech event during 2025 SF Tech Week with 1,000+ registrations. Our conversation with OpenAI’s former Head of Sales Aliisa Rosenthal—who scaled OpenAI from $10M to $10B in revenue—revealed something critical for founders: financial services enterprises are ready to buy from early-stage startups, right now.
You can find key takeaways below and the full recording of the event at the end.
Financial Services Moved Faster Than Anyone Expected
Everyone assumed tech companies would lead AI adoption. They were wrong.
Three industries surprised the OpenAI team by becoming early, aggressive adopters: #1 financial services, #2 healthcare, and #3 manufacturing & industrials. Financial services weren’t just first—they moved with unexpected speed and conviction.
Why?
Content-heavy workflows made them perfect for AI from day one. Investment memos, research reports, document analysis—these weren’t nice-to-haves, they were immediate use cases with clear ROI.
The economics were obvious. Offsetting even 10-20% of a banker’s or research analyst’s time delivers massive returns when you’re paying $200K+ in comp.
Client demand created competitive pressure. Wealth management clients started asking “How are you using AI to optimize my portfolio?” Firms that couldn’t answer lost business.
Enterprises Were Ready Before the Product Was
Aliisa’s first meeting at OpenAI was with Morgan Stanley in June 2022. At the time, OpenAI had zero enterprise certifications, no compliance framework, and was selling a $60,000 “AI innovation license” that literally gave you one hour of conversation about AI.
Morgan Stanley signed anyway. They built a wealth manager chatbot using retrieval technology before ChatGPT even existed.
Carlyle showed even stronger conviction. Before ChatGPT Enterprise had pricing figured out, they told OpenAI: “Just send me a contract. I’ll sign it.”
When the OpenAI team asked what Carlyle was building, they showed deal diligence automation, Black-Scholes analysis, and sophisticated quantitative use cases. “It blew our minds,” Aliisa recalls. “We weren’t even thinking about ChatGPT as that sophisticated.”
Insight for founders: your customers will often see use cases you never imagined. Even the largest enterprises in the world are ready to design and build the product together with you in your earliest days. The $20M Enterprise Budget That Can’t Ship
Here’s the uncomfortable truth: even enterprises with $20M AI budgets and direct Slack access to OpenAI engineers struggle to implement meaningful AI solutions.
The gap is real. Most companies drastically underestimate how hard it is to work with APIs. They lack developers who understand model strengths and weaknesses, can manage latency, run evals, and implement retrieval systems. Their procurement processes are built for buying seats of Salesforce, not for iterating on AI applications.
CIOs have massive budgets to spend, genuine risk tolerance, and big dreams. What’s missing? Service providers who can actually fill that gap.
Insight for founders: this is your opening. Enterprises will access AI primarily through application layers, not by building in-house. They need you to be that missing service provider. And unlike previous technology waves where startups had to “land and expand” from SMBs, enterprises are ready to be your first customer.Two Waves of Disruption Ahead
During Aliisa’s time at OpenAI, the company’s own accounting team cut a 4-day Excel tax calculation process down to 10 seconds using ChatGPT, with verified accuracy. The impact was so profound that the employee who built it left to help other companies automate accounting full-time.
This is exactly the kind of transformation happening across financial services right now. We see two distinct waves:
Wave 1: Vertical AI displaces incumbent solutions. Startups are building AI copilots that simply do the job better than 40-year-old tools like Bloomberg Terminal, FactSet, or manual Excel processes. Better efficiency, better accuracy, 10x better user experience.
Wave 2: Complete workflow re-imagination. This is where it gets most interesting. Instead of just making the old process faster, founders will rethink the entire flow. Imagine one natural language interface managing all your taxes, investments, 401(k)s, and financial documents in a single place. Not automation—reinvention. The opportunity spans loan origination, insurance underwriting, portfolio management, risk analytics, compliance, legal doc review and much more.
Advice for Founders: Be Audacious
Aliisa’s advice to founders building in this space:
“My biggest advice is just be audacious and dream big, and don’t feel limited by the current constraints of the environment that you’re in. We’re going to see things change so dramatically over the next few years.
Instead of replacing an existing flow, rethink the entire flow, rethink the entire way that users might interact with or deal with the problem that you’re out to solve.
Lastly, just be audacious. Be bold. Make big bets, and don’t be afraid to sell to financial services because they’re along for the ride with you.”
We couldn’t agree more. The enterprises are ready. The budgets exist. The question isn’t whether this disruption will happen — it’s who will build it.
At Elsa Capital, we’re excited about backing the founders audacious enough to rethink the entire financial and professional services flow, not just automate the existing one.Event Recording
You can watch the full conversation here: View recording
If you’re building vertical AI for financial or professional services, we’d love to hear from you. Reach out here or directly at sarah@elsacap.com.
Follow us on LinkedIn or subscribe to our Substack to stay updated on future events.
The point that stuck with me most was that enterprises were often ready before the products were. I think many founders underestimate how much conviction early customers can have when the pain is real.