The Search for the AI-Native Firm (David Peterson, Angular Ventures)
| Author | David Peterson (Angular Ventures, The Angle) |
|---|---|
| Published | November 17, 2026 |
| Format | Newsletter |
| Original | Read the original |
AI adoption is going to be fast and AI transformation slow; every current model (SaaS 2.0, AI roll-ups, AI-native services firms) fits AI into old structures, and the big gains will come from organizations redesigned around what AI makes possible.
The Search for the AI-Native Firm (David Peterson, Angular Ventures)
Key claims
- Thesis for the next decade: adoption (everyone chatting with ChatGPT or Claude) is fast; transformation (redesigning the organization assuming AI does much of the work) is slow and needs new products, business models and org structures.
- Today’s AI software winners are “SaaS 2.0.” Horizontal (Lovable, Gamma) or vertical (Harvey and Legora, Rogo and Hebbia), they are opinionated harnesses on models the way SaaS 1.0 was opinionated interfaces on databases. They still sell per seat per month, and customers are pushing back as build-versus-buy shifts.
- The “SaaSpocalypse is real”: nobody will rebuild Salesforce, but nobody needs to, because a company starting fresh would build the 1% it uses, and frontier coding models can already do that. Outcome pricing is the escape hatch, but outcomes are hard to define and verify, and defining one makes it easier for customers to shop around.
- Selling the work, not the software (Sarah Tavel’s “Sell Work, Not Software”, Sequoia’s “Services: The New Software”) has two forms. AI Rollups buy services firms and deploy AI, but run on debt and five-to-seven-year exits, a short clock if transformation is slow, and may only layer AI onto an old structure. AI-native services firms (a thesis championed by Emergence) design from scratch but end up running a services business that still needs sales, account management and edge-case staff, and may lose engineers who joined a tech company.
- Peterson’s alternative: ask how you would build a new organization today for a given problem. Three questions follow. Where is the bottleneck (Goldratt’s The Goal and the Theory of Constraints: automate 90% of admin work and growth can still be capped by licensed professionals)? Where are the economies of scale (Ronald Coase’s The Nature of the Firm, with Alex Imas’s paper Organizing Intelligence on AI changing both production and coordination costs)? And what compounds (does each customer improve the product, the data or the cost base)?
- The historical analogy is Electrification: factories first swapped an electric motor in for the steam engine and kept the line shafts and layout; it took decades to put a motor on each machine and lay out the floor around the flow of work.
Notable quotes
Here’s my entire thesis on the next 10 years: AI adoption is going to be fast. AI transformation is going to be slow.
SaaS 1.0 products were opinionated interfaces on top of databases. SaaS 2.0 products are opinionated harnesses on top of models.
The standard pushback on vibe coding in the enterprise has been “nobody will rebuild Salesforce.” And that’s true. Because nobody would build that gigantic piece of garbage.
So we have a technology that’s increasingly capable of doing the work. But we’re stuck trying to figure out how to sell it as software. Maybe the problem isn’t how we’re charging for the software. Maybe it’s that we’re still selling software at all?
The question isn’t how much work AI can automate, but whether it removes the constraint. And if it does, where does the bottleneck move next?
The lesson is obvious enough to slap us in the face: the opportunity afforded by electricity wasn’t to replace the steam engine, it was to redesign the factory around what electricity made possible.
Ultimately, what we’re doing is mistaking adoption for transformation.
How it connects
- AI-Native Firms: the concept this essay is about; Peterson’s three questions (bottleneck, scale, compounding) are a usable test for whether a company is one.
- Services as Software and AI Rollups: the two “sell the work” models he critiques.
- Theory of Constraints, The Nature of the Firm / Ronald Coase, Electrification: the three frameworks he leans on.
- The issue’s “Worth Reading” roundup summarises Kyle’s own The McDonaldization of Venture Capital (“Harrison argues that venture is increasingly dominated by massive asset managers…”) right after a Jerry Neumann piece on the death of venture.
- Workers Are Not Going to Be OK (Azeem Azhar, NYT): the macro version of the same shift, AI budgets moving out of IT and into the departments that hire.
- Saved via Absolutely fantastic piece by my… (tweet) from Gil Dibner, Peterson’s partner.