According to Fortune, shortly after Ashok Srivastava became Intuit’s chief AI officer last summer, he merged the AI unit with the “futures” team to form “Intuit Foresight,” a group of a few hundred employees. This team built a foundational AI model used in a new accounting agent, which debuted in July as part of what CEO Sasan Goodarzi called the company’s “most significant launch.” That agent now saves customers an average of 12 hours per month and categorizes transactions with over 90% accuracy. In November, Intuit’s AI products drove better-than-expected fiscal first-quarter results across TurboTax, QuickBooks, and Credit Karma. Last month, the company also signed a $100 million multiyear contract with OpenAI. Since July, over two million customers have engaged with Intuit’s AI agents, with repeat engagement above 80%.
The Strategy: Stitching Research to Revenue
Here’s the thing about Intuit’s move: it’s not just about having a fancy AI lab. The entire mandate of Intuit Foresight is to “stitch together” pure research with practical applications that developers can actually deploy. That’s a business strategy, not just an R&D project. The payoff is clear in the numbers they’re sharing: 40% faster coding internally, 39% more code per developer, and 3,500 generative AI use cases created in a year. They’re turning AI into a direct lever for productivity and, ultimately, revenue growth. The $100M OpenAI deal is another smart tactical move—bringing their platform capabilities into ChatGPT is a huge distribution play. They’re not just building AI; they’re embedding it everywhere their customers already are.
A Cautionary Tale for Other Companies
But here’s where it gets really interesting. Srivastava explicitly warns other large organizations not to blindly copy this “lab” model. His reasoning is brutally practical. He says if a company doesn’t have its data in order, a solid AI platform, and the right talent, creating a team like Intuit Foresight is “premature.” He points out they’ve been working on this for nearly a decade. That’s a huge admission. Basically, the flashy team structure is the visible tip of a massive, long-term infrastructure iceberg. A lot of companies see the AI hype and think they need a Chief AI Officer and a dedicated team tomorrow. Srivastava is saying, “Slow down. You probably don’t have the foundation we spent ten years building.” It’s a rare moment of sober advice in an industry fueled by FOMO.
The Real War is for Talent
And that foundation isn’t just software and data—it’s people. Srivastava acknowledges the “increasingly expensive war for AI talent,” but then makes a fascinating point. He says the technologists he’s recruiting aren’t only motivated by money (even at a financial tech company!). He sells the mission of serving small businesses and is “shocked” by the unprecedented number of applicants. That tells you something about the market. Yes, salaries have to be competitive, but for top-tier AI scientists and engineers, the chance to work on models that directly impact millions of small businesses—and see tangible results like saving 12 hours of drudgery—is a powerful draw. It’s a talent strategy that goes beyond the paycheck, which you need when everyone is throwing money at the same problem.
So What’s Next? Reasoning and Forecasting
Looking ahead, Srivastava says the priority is moving beyond today’s generative AI tools to models that excel at “reasoning” and “forecasting and scenario planning” specialized for finance. That’s the next frontier. The current agents handle categorization, but the real value for a company like Intuit is predictive insights: cash flow forecasting, tax scenario modeling, personalized financial advice. That’s where the 90% accuracy needs to get closer to 99.999%, and where latency becomes even more critical. The progress so far, from faster coding to millions of customer engagements, is just the setup. The real transformation—and the next phase of their competitive moat—will be building AI that doesn’t just automate tasks but actively reasons about complex financial futures for its users. Not every company can, or should, try to follow that path.
