AI as Feature vs AI as OS
person concept tool org talk claim — click a node to jump to its page; hover an arrow for the relation
The distinction between "AI as Feature" and "AI as OS" contrasts two architectural philosophies: using AI as an add-on capability within existing software versus rebuilding software around AI as the foundational operating layer. The concept is closely tied to the idea that AI-native software should look like an agent wrapping deterministic tools, not deterministic software wrapping an AI.
Core Argument
Pete Koomen advocates strongly for the "AI as OS" approach, arguing that the key shift for companies is to stop treating AI as a co-pilot bolted onto existing workflows and instead use it as "the building layer for everything." (Inside YC's AI Playbook, 0:25)
The Anti-Pattern: AI as Feature
The Horseless Carriages concept is invoked as a criticism of the "AI as Feature" approach — the mistake of mapping new technology onto old paradigms. The pattern being criticized is that of "companies building software and adding AI features by sort of slotting a little bit of AI inside of a lot of software." (Inside YC's AI Playbook, 32:54) This framing implies that incrementally adding AI to conventional software architectures is analogous to putting a motor on a horse-drawn carriage — preserving the wrong structural assumptions.
Architectural Implication
The positive vision of "AI as OS" is defined structurally: the correct inversion is an agent at the center orchestrating deterministic tools, rather than a deterministic software system that occasionally calls out to an AI subsystem. This makes AI the control flow and decision-making substrate, with conventional code serving as the toolkit it wields. (Inside YC's AI Playbook, 0:25)