<- all tokdocs

What Is an AI-Native Startup? Tim Chen on Moving Past AI Coding

By , founder of Mavric and Apso

View on TikTok ->

An AI-native startup reorganizes its product, pricing, go-to-market, team and hiring around what AI makes possible, rather than only using AI to write code faster. That is the argument in a TikTok from venture investor Tim Chen (@timchenvc). He describes a two-year-old portfolio company that banned hand-written code and then found that faster coding was only the first layer of a full company overhaul.

What does AI-native mean for a startup?

Chen draws a line between two kinds of company:

  • AI-assisted. Developers use Copilot or a coding agent to write code faster. The company sells the same product and works the same way.
  • AI-native. The whole company is restructured around what AI makes possible and what it makes obsolete, including the product itself.

His point is that many young startups are in the first group and believe they are in the second. They are "still selling the same product, doing the same thing," and only using AI to write the code behind it. In his view, that is no longer enough.

Why did one startup ban writing code by hand?

The portfolio company Chen describes has gone "fully AI-pilled." Its first rule: no one on the team may write code manually, except in narrowly scoped sections. Everyone uses AI coding agents.

The rule goes beyond individual habits. The whole team is expected to build evaluation harnesses and testing infrastructure so the agents produce reliable output across the codebase. In this model, engineering effort moves from writing features to building the checks that keep agent-written code trustworthy. Chen says he sees this happening across the industry, not just at one company.

What happens to product strategy when software feels free to build?

Chen's core concern is that software increasingly feels free to build. Language models and coding agents have improved enough that the cost of shipping a feature keeps falling. If that trend continues for another six months, he argues, a startup whose only edge is "we built a good product" will struggle, because anyone can now build a good product.

That raises a practical problem. If your customers can build your features with their own agents, your value has to come from somewhere else.

What should an AI-native startup sell?

Chen suggests the product should help users get more out of their own AI tools, in areas where coding agents fall short on their own. He names three examples:

  • Distribution, reaching customers in ways an agent cannot.
  • Domain knowledge, the detailed understanding of a field that raw AI tools lack.
  • Curated context, information selected and organized so that a user's agents produce better results.

In this framing, the product makes the customer's own agents smarter rather than competing with them.

Why are young startups facing an existential crisis?

Chen reports seeing the same pattern across his portfolio. Companies founded only one or two years ago are already facing an existential reckoning. The teams that treat AI as a code-writing productivity tool have not worked through what AI-native means for the rest of the business.

He lists what has to change:

  • Product pricing.
  • The go-to-market motion.
  • The sales approach.
  • Team composition.
  • Who the company hires.

He describes it as a full organizational transformation, and says he is watching it happen across the companies he works with.

How to tell whether your company is AI-native

Chen's framing suggests a few questions a founder can ask:

  • If a customer could build your core features with their own coding agent, what would they still pay you for?
  • Is your engineering team building evaluation and testing infrastructure for agents, or only using agents to type faster?
  • Have your pricing, sales motion and hiring plan changed since your team started using AI, or only your code editor?

A company that answers "nothing," "only typing faster" and "only the editor" is, in Chen's terms, AI-assisted.

Bottom line

Chen's argument is that coding with agents is now the baseline, so it cannot be the strategy. The startup he describes banned hand-written code, put its engineers on evaluation infrastructure, and is rethinking its product around distribution, domain knowledge and curated context. Founders whose only AI change is faster coding should expect to revisit pricing, sales and hiring too.

Sources

Published June 11, 2026. Based on a TikTok by @timchenvc. Full video notes (markdown)