In a recent deep dive with Sequoia Capital partner Pat Grady, he shared surprising insights about what separates winning AI companies from the rest, and challenged conventional wisdom about AI moats, pricing models, and what investors truly value.
Our favorite takeaways:
Building AI companies is just building a company. It’s 95% the same and people problems still dominate
Trust is the critical design pattern most AI companies miss. Users need to see how you arrive at your results
Most AI products achieve 80% functionality quickly, but the final 20% takes 5-10x longer and is what builds actual trust.
The greatest moat in AI isn't data or tech - it's founders with relentless execution.
Pat also added some extra wisdom that we appreciate:
The "data flywheel" appears in 100% of AI pitches but only 1% of companies actually demonstrate it works - Pat demands evidence, not theory
AI pricing will standardize around outcome-based models with huge variation - the most successful companies think about both "input" (work done) and "output" (value created)
For investors, negative gross margins are acceptable in early AI companies because token costs are dropping 99% and multi-tenancy is becoming more accepted
Domain-specific AI products that build real trust can carve out defensible positions against foundation model providers in vertical markets
The most successful AI companies avoid "vibe revenue" (temporary excitement) by focusing on engagement and retention using consumer internet metrics even for B2B products
What's your experience with AI products and pricing models? Have you found user trust to be the limiting factor? Share your thoughts below 👇
Share this post