This week my guest is Rohan Parikh, founder and CEO of Keye, an AI-powered platform revolutionizing due diligence for private equity professionals. Rohan brings a fascinating perspective shaped by growing up in a family of entrepreneurs in Mumbai and spending five years climbing from analyst to director at a French investment bank in New York. We dive deep into why deterministic software is non-negotiable in private equity, where a single investment can represent 5-10% of an entire fund and career-defining decisions hinge on absolute precision. Rohan shares his journey from building a B2B information services company to pivoting Keye into an AI solution that processes millions of rows of Excel data to uncover the true health of businesses during diligence.

Key Takeaways:

• The determinism imperative: In private equity, where single investments can make or break a fund’s returns, non-deterministic AI outputs are unacceptable—investors need to understand exactly how every number was calculated
• Excel’s limitations exposed: PE firms routinely handle data files with millions of rows that Excel literally cannot open, creating a massive bottleneck in the due diligence process
• The $50 billion opportunity: Due diligence spending in private equity exceeds $50 billion annually, representing a massive market for workflow transformation
• Start with the problem: Rohan’s advice to first-time founders is to focus on solving a specific problem rather than starting with a technology and looking for applications
• Building for the long game: Keye isn’t trying to be “AI for finance”—they’re laser-focused on owning the entire private equity workflow from sourcing to portfolio management

What strikes me most about Rohan is his clarity of vision. While the entire world is rushing to slap AI onto everything, he’s building with the discipline of someone who understands that in private equity, being 95% accurate isn’t good enough—you need 100%. That constraint has forced Keye to think differently about how they architect their system, combining deterministic computation with AI to deliver both accuracy and insight. It’s a masterclass in understanding your customer’s actual needs rather than what you think they need.