Skip to content
Y CombinatorY Combinator

How to Build an AI-Native Services Company

Some of the biggest companies of the next decade won't be software businesses. They'll be services companies like insurance carriers, law firms, and tax practices rebuilt from scratch with AI doing most of the work. In this episode of Startup School, YC Visiting Partner Charlie Warren walks through the playbook for building AI native services companies, covering how to pick a market with the right traits, why variance kills these businesses faster than anything else, and the P&L math that’ll transform your business model. Chapters: 00:00 — Intro to AI Services Companies 01:01 — Picking the Right Market 02:55 — Markets YC Likes Right Now 03:43 — The Sam Altman Test 04:35 — The Right Founding Team 05:28 — Building the Product 06:19 — Variance Is the Existential Problem 07:08 — The Early Demand Trap 07:53 — How to Price AI Services 08:41 — The P&L Walkthrough 09:33 — AI Operating Leverage 10:27 — Don't Buy Your Way In Apply to Y Combinator: https://www.ycombinator.com/apply Work at a startup: https://www.ycombinator.com/jobs If you're in the Bay Area and Charlie's video resonated with you, come join us in San Francisco on 6/22. We're hosting a founder panel on AI-native services across industries, followed by a happy hour with the panelists, YC partners, and a room full of engineers, researchers, and builders thinking about this space. Apply to attend: https://events.ycombinator.com/3EKUFJ6V7

Jun 3, 202611mWatch on YouTube ↗

At a glance

WHAT IT’S REALLY ABOUT

YC playbook for building AI-native services companies from scratch

  1. AI-native services deliver outcomes (not tools), targeting trillion-dollar sectors like tax, insurance, law, and regulated healthcare niches.
  2. The best markets share traits like existing outsourced spend (low trust), mostly-automatable task steps (low judgment), high complexity, and often regulation as a moat.
  3. Winning teams combine domain fluency, frontier-model fluency, and operational rigor because the “product” is the service operation.
  4. Operational variance is the existential risk: inconsistent outputs destroy trust faster than slower speed or higher price.
  5. Economics hinge on AI operating leverage—COGS (models/hosting/humans) must fall over time to move from ~30% services margins toward 50%+ software-like margins—while avoiding traps like overloading on pilots or buying legacy firms for instant revenue.

IDEAS WORTH REMEMBERING

5 ideas

Pick markets where spend already exists and behavior change is minimal.

Target “low trust” work that’s already outsourced, so you replace a vendor and sell the outcome rather than persuading customers to adopt a new internal workflow.

Design for mostly-automatable workflows with limited human judgment checkpoints.

If every step requires expert judgment, headcount scales linearly with revenue; you need work that can be decomposed so humans focus on a few high-leverage decisions.

Use the “Sam Altman test” to avoid being commoditized by better models.

Choose services where improving models make your offering stronger (faster/cheaper/better delivery) rather than making the model itself the entire product customers buy directly.

Treat operational metrics as core product metrics.

Throughput, cycle time, and bottlenecks should be tracked like DAUs because the human-facing operation is the customer experience and the software’s job is to amplify it.

Variance is a bigger churn driver than speed or price.

Non-uniform outputs erode trust; customers will tolerate slightly slower or pricier service, but inconsistency signals unreliability and triggers replacement.

WORDS WORTH SAVING

5 quotes

Some of the biggest companies of the next decade won't be software businesses at all. They'll be services companies like insurance carriers and law firms rebuilt from scratch with AI doing most of the work.

Unknown

You should ask yourself, as the models get better, does your service get stronger, or does the model itself commoditize you? You wanna be in the first camp.

Unknown

You have to bleed credibility.

Unknown

Variance Is the Existential Problem here.

Unknown

It's easy to sign up a lot of pilot customers when you're just starting out and have nothing, but it can quickly overwhelm your ability to serve them, and you won't be able to build the product to scale. You'll be stuck using humans. It is a literal trap.

Unknown

AI-native services vs copilotsMarket selection traits (low trust, low judgment, high intelligence threshold, regulation)The Sam Altman test (benefit vs commoditization)Founder profile: domain, model, and ops fluencyOperations as product: throughput, cycle time, SOPsVariance management and trustPilots, pricing, P&L, and AI operating leverageWhy buying legacy service firms usually fails

High quality AI-generated summary created from speaker-labeled transcript.

Get more out of YouTube videos.

High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.