Ravi Gupta announces new company while staying at Sequoia, Pat Grady outlines why now is the best time to start an AI application-layer company

Summary

Ravi Gupta is launching a new company while retaining a partner-level role at Sequoia Capital, stepping back from his general partner responsibilities but staying on existing boards. The co-founder is undisclosed but described in emphatic terms as substantially more accomplished than Gupta himself, enough that his children, initially skeptical about the venture, offered their combined $750 in birthday savings as seed capital.

Gupta's departure from day-to-day GP work reflects a deliberate bet on AI timing rather than a specific idea. His framing, drawn from a piece he wrote roughly a year ago titled "AI or Die," is that competitive advantage now belongs entirely to fast-moving, reactive organizations willing to fight for relevance daily. He explicitly rejects the narrative that the best founding windows have passed, calling that line of thinking low-agency.

Pat Grady makes the affirmative case for the application layer with more structural precision. He identifies three inflection points in AI capability: November 2022 with ChatGPT and pre-training; fall 2024 with OpenAI's o1 and reasoning models; and the most recent weeks, where Claude's long-horizon agent behavior, specifically Opus 4.5 with extended coding tasks, demonstrated what persistent autonomous agents can accomplish. His argument is that application-layer companies can now deploy always-on agents running multiple parallel instances around the clock, a fundamentally different value proposition than the chatbot interactions of two to three years ago.

Sequoia's core investment thesis holds that value accrues at the application layer, not the foundation model layer, a view Grady says is increasingly validated even by the economics of the labs themselves. On durability, both Gupta and Grady converge on two signals worth building around: network effects, which they argue AI cannot easily replicate on demand, and the quality of founding teams, which they treat as the primary variable that determines whether a company navigates unpredictable technical shifts.

On the leadership transition at Sequoia, Grady and Gupta describe the steward role under Alfred Lin as lighter in management load than outsiders assume. The firm runs with fewer than 20 general partners, and the steward function is characterized as thought-partnership rather than active oversight, with investment performance remaining the only metric that matters for institutional credibility.

Both push back on the idea that AI lacks a Jobs-style public communicator. Gupta surfaces an anecdote attributed to Alfred Lin about Lin's father working as a human calculator before technology freed him to build accounting systems, framing AI as a tool whose moral valence depends entirely on how it is deployed. The broader concern raised is that public discourse on AI is now roughly split between grounded product analysis and speculative science fiction, a ratio that Grady and Gupta argue makes customer feedback and real-world product signals more important than ever as a navigational anchor.