Matt Biilmann, CEO and founder of Netlify and creator of the Jamstack, joins A16Z general partner Martin Casado to discuss the dramatic transformation happening in web development. Biilmann coined the term 'Agent Experience' (AX) and has watched Netlify evolve from serving 17 million professional JavaScript developers to potentially reaching 3 billion spreadsheet users as AI democratizes software creation.

The conversation covers Netlify's explosive growth from 3,000 to 16,000 daily signups, driven largely by non-developers using AI tools to build websites. They explore how the fundamental skills of being a developer are shifting from syntax mastery to systems thinking, and why frameworks like React and Next.js represent 'wasted knowledge' that AI can now handle.

Biilmann shares his personal journey of learning to code again without writing code directly, using AI agents to submit pull requests and build products. The discussion examines three types of Agent Experience, the economics of token-based pricing, and why the 'dead web' theory misses the creative explosion happening as barriers to software creation collapse.

The Great Developer Audience Expansion

Netlify's addressable market exploded from 17 million professional JavaScript developers to potentially 3 billion people who can use spreadsheets, as AI removes coding barriers.

Daily signups jumped from 3,000 to 16,000, with only 4% coming from direct AI tool integrations like Bolt.new - the rest represents organic growth from people using AI to build websites.

New user personas now include marketers, designers, and product managers who previously couldn't write code, fundamentally changing who builds software.

Three Layers of Agent Experience Architecture

Netlify's AX focuses on making their CLI and documentation accessible to AI agents, using techniques like content negotiation to serve markdown instead of HTML to reduce token usage.

Customer AX helps Netlify users build agent-accessible sites, like integrating with Stripe's new ChatGPT payment options for e-commerce customers.

Industry AX requires developing protocols and standards across the web ecosystem to enable seamless agent interactions with websites and services.

The Economics of AI-Driven Development

Biilmann accidentally spent $10,000 letting an AI agent build a game autonomously for hours, illustrating the unpredictable token costs of agent-driven development.

The industry is shifting from recurring subscriptions to usage-based pricing, with companies seeking outcome-based models rather than pure token consumption.

25% of users who encounter build failures immediately copy error logs to LLMs, showing how AI has become integral to the debugging workflow.

Redefining What Makes a Developer

Core developer skills are shifting from syntax mastery to systems thinking, clarity of thought, and understanding user needs as AI handles code generation.

Framework knowledge like React and Next.js represents 'wasted knowledge' - arbitrary design decisions taking up brain space that AI can now handle.

Software development will become a universal skill like writing, with professional developers remaining but basic coding competency expected across many roles.

The CEO Coding Renaissance

Biilmann made it his mission to become 'an equally good developer without writing any code myself' and now regularly submits pull requests using AI agents.

Founder CEOs are returning to coding through AI tools because they've already learned to delegate and work at higher abstraction levels.

AI enables asynchronous development where executives can check progress between meetings, giving feedback to agents running autonomous coding tasks.

The Creative Web Explosion vs Dead Web Theory

Biilmann has visited more new websites in six months than the previous five years, contradicting the 'dead web' theory with an explosion of creative projects.

People are building 'really cool WebGL games and so on that they could just never manage to build before' as AI removes technical barriers to creative expression.

The constraint that only ~100 million developers could build software for 8 billion people is disappearing, promising an 'unimaginable amount of more software.'

The Future of Web Consumption

Users develop strong preferences for their specific AI assistants (ChatGPT vs Claude) and want to filter all web interactions through their chosen agent rather than visiting individual company websites.

The web's consumption layer is shifting from UI-based to conversational, with voice interactions potentially preferred over traditional screen readers for accessibility.

Browsers will evolve dramatically as the original concept of 'user agent' becomes literal, with AI agents as core users of every web product.