Descriptions:
A Y Combinator partner lays out a detailed playbook for founders building what YC calls AI-native services companies — businesses that deliver outcomes directly to customers rather than selling copilot software for customers to operate themselves. The talk argues that advances in frontier models have unlocked a fundamentally new company archetype in markets like tax, audit, insurance, law, and parts of healthcare, with combined TAMs in the trillions of dollars.
The presentation identifies four criteria that distinguish high-fit markets: low trust (work is already outsourced, so no behavior change is required from the buyer), low judgment at the task level (most steps are automatable with judgment concentrated in a few human-in-the-loop nodes), a high intelligence threshold (genuinely hard enough that models plus humans are necessary), and regulatory moats (compliance requirements that raise the floor for all competitors). YC portfolio companies Panacea — which pairs experienced FDA consultants with an AI platform to deliver regulatory approvals for biotech and medtech — and General Legal, an AI-native law firm founded by veterans of Cooley, Fenwick, and Casetext, are cited as concrete examples of the model working in practice.
The playbook also covers founding team composition (domain fluency, model fluency, and operational rigor are the three non-negotiable attributes), product philosophy (the human is the customer-facing interface; the software scales the human), the ‘early demand trap’ of signing too many pilots before the product can handle them, and why output variance — not speed or cost — is the metric that determines whether customers stay or churn. The talk closes with a brief framework for evaluating whether to acquire an existing services business rather than build from scratch.
📺 Source: Y Combinator · Published June 03, 2026
🏷️ Format: Opinion Editorial
