Jack Raines on Slow Ventures' AI thesis: vertical AI is the new vertical SaaS, not AI-for-AI's sake
Sep 10, 2025 with Jack Raines
Key Points
- Slow Ventures sees AI matured past hype: the pattern at YC Demo Day is vertical software companies using AI as table stakes, not AI-native startups pitching AI itself.
- The founder archetype Slow backs pairs an AI technologist with a domain expert who spent 20 years in the vertical, not an AI generalist retrofitting a problem.
- Foundation model companies and hyperscalers will consolidate infrastructure; durable businesses live far up the application stack, grounded in specific domains where commoditized models can't erode the moat.
Summary
Jack Raines, investor at Slow Ventures, views the current YC batch as evidence that AI has moved beyond hype. Demo Day featured vertical software companies using AI as a tool, the way earlier generations used cloud or mobile. The shift in investor questions is real: not "do you have an AI strategy" but "what are you actually building."
Slow Ventures manages around $1 billion and writes $1–3 million checks across stages and verticals. Raines is direct that the firm respects AI used well, not AI as a differentiator in itself. If a company's only edge is AI, that's a problem. It's table stakes, not a moat.
The founder pattern that catches Slow's attention pairs an AI technologist with a genuine domain expert who has spent 20 years inside the vertical. That combination, rather than an AI generalist solving a problem retroactively, is where Raines sees durable businesses. The parallel to vertical SaaS is intentional: those winners went deep into specific industries with purpose-built workflows, not by building better generic software.
A concrete example from Demo Day is a company running the 311 line for Florida municipalities and fielding alligator calls. Niche, specific traction, real domain knowledge. That's the model.
Foundation model companies and hyperscalers will continue consolidating the infrastructure layer, with Microsoft acquiring companies that resell compute. The companies worth backing are those high enough on the application stack and grounded enough in a specific domain that commodity models don't destroy the moat.