Everyone Can Cook: Guillermo Rauch on AI-Native Product Building, Full-Stack Generalists, and the Future of Software

Guillermo Rauch, founder of Vercel and creator of Next.js, recently sat down with Lenny Rachitsky and delivered one of the most thought-provoking conversations about the future of building software in the age of AI.

This isn’t just another "AI will change everything" take. It’s a grounded, real-world map for what product builders should do now — and what roles might look like in 2, 5, or 10 years.

Here are the most important lessons I took away.

1. 💬 “Code Last”: A New Paradigm for Building Products

The biggest shift: we’re moving from “code first” to “intent first.”

In the traditional model, you:

1. Define a problem.
2. Write specs.
3. Wait for design, then code.

In the v0 model:

1. You prompt.
2. You prototype instantly.
3. You iterate through conversation.

“The Git commit used to come after you wrote the code. With v0, you start with the intent, and it gives you the code.”

Implications:

• The PM’s role now starts at creation, not documentation.
• Design-to-dev handoff becomes obsolete in many cases.
• Founders can ship a v1 before hiring a team.

This doesn’t mean code is dead — but it moves to the background as infrastructure.

“It’s not ‘no code.’ It’s ‘code last.’ You start with what you want, not how you get there.”

2. 🧑‍🚀 Everyone Can Ship — Not Just Engineers

This might be the most empowering message for non-technical builders.

“Some of our best v0 creations didn’t come from engineering. They came from marketing, product, even sales.”

With v0, any team member can:

• Build a live feature.
• Iterate with real logic.
• Push to production.

Guillermo calls this “the rise of the full-stack generalist.” The barriers are gone — but the judgment still matters.

Example:

• Vercel PMs spec and build new flows directly in v0.
• A marketing team member shipped an animation explaining Vercel infra better than any diagram before.
• A user built a live flight radar app mid-flight on spotty Japan Airlines Wi-Fi — in < 2 hours.

“We have 600 people at Vercel. It used to be 150 engineers. Now? It’s like we have 600 engineers.”

3. 📈 TAM Expansion: From 5M Devs → 100M Builders

Guillermo’s product thesis for v0 is razor sharp: expand the total addressable market of creators.

“There might be 5 million React devs. Maybe 20 million JavaScript devs. But there are 100 million people who want to build products.”

How’d he get that number?

• Slack has ~100M MAUs.
• Most of them are in product-facing roles.
• They ideate, write specs, talk to customers — and now they can also build.

Strategic framing:

• v0 = TAM unlock for Vercel, just like Shopify unlocked ecommerce entrepreneurs.
• The power of v0 is distribution + empowerment: every team member becomes a shipper.
• PMs are no longer constrained by bandwidth from other teams.

“What if you could just yap into the computer — and it worked?”

4. 🛠️ Skills for the AI Era: What Still Matters (and What Doesn’t)

Guillermo doesn’t sugarcoat it — many software jobs are translation tasks, and those are going away.

“I used to hire people whose only job was turning a Figma into CSS. That task is gone.”

But some skills are more valuable than ever:

• Systems thinking: Knowing how software really works under the hood.
• Math: The foundation for logic and modeling.
• Eloquence: The ability to prompt precisely, describe intent, and think symbolically.
• Taste: The ability to judge what’s good and steer AI outputs accordingly.

“Knowing the tokens — layout, flexbox, Tailwind — helps you influence the model better.”

He adds:

• His latest website, previously built with Next.js manually, now takes just 2 prompts in v0.
• And the generated version is more accessible than the one he built by hand.

If you're early in your career:

• Learn symbolic language (HTML, CSS, React, SQL).
• Learn how to describe what you want clearly.
• Understand systems — not just tools.

“I’m not saying don’t learn to code. I’m saying learn what the code means.”

5. 🎨 Default Beautiful: Why Design Is No Longer a Bottleneck

One of the most surprising — and practical — revelations from the conversation: v0 generates production-grade design by default.

It uses tools like:

• shadcn/ui for components (used by Grok, Perplexity, etc.)
• Tailwind CSS
• Next.js for routing and structure

Guillermo compares it to working with a designer:

• You can say: “Make it look like Charles Schwab” or “Apply sepia tone like Semaphore”
• You can paste a screenshot and say: “Give me a cleaner version of this”
• You can refine with language like: “Make this more serious,” or even just “Try something else”

“One of the hardest problems in computer science is still centering a div. But v0 just does it.”

Why this matters for PMs and Founders:

• You no longer need a dedicated designer to create polished UIs.
• You can now go from idea → user-testable product in hours, not weeks.
• You can iterate visually with simple feedback.

This doesn't make design obsolete — it makes taste more important.

6. 👁 Building Taste Through Exposure Hours

If design is democratized, then taste becomes your edge.

“Taste is not some innate thing. It’s built through what we call ‘exposure hours.’”

Guillermo's advice:

• Use more great apps.
• Watch people use your product (and others).
• Show your work. Feel the pain. Ship again.

At Vercel, it’s built into the culture:

• Friday “Demo Days” where anyone — PMs, designers, engineers — can show off what they made in v0.
• Frequent dogfooding of both Vercel and v0 itself.
• Color-coded calendars to ensure a balance of team syncs and customer-facing time.

“We tend to overrate how well our product works. You have to expose yourself to the pain of reality.”

Practical tip: build the habit of watching others use your work, even if it’s just shadowing a user for 10 minutes.

7. 🌱 Community-Led Product Loops (The GitHub of AI Building)

v0’s strongest moat might not be the tech — it’s the community flywheel it’s creating.

In under a month:

• v0 Community saw 20,000+ public submissions
• Projects have thousands of forks
• Builders remix, refine, and reshare — like GitHub for no-code AI apps

“Instead of starting from scratch, you start from someone else’s app and prompt it into your own.”

Strategic Insight:

• This is social product building, not just social coding.
• PMs can fork, tweak, and test ideas without any devs.
• New best practices, patterns, and UI styles emerge organically.

And because it’s built on open-source components (Next.js, shadcn), the community outputs are already “enterprise-grade.”

8. 🏭 The v0 of Everything: Vertical AI as a Playbook

One of the most exciting opportunities Guillermo points to: building “v0 for X.”

Examples already emerging:

• chatprd.ai: the v0 for writing Product Requirement Docs
• OpenEvidence: AI for doctors
• GetGC.ai: Legal assistant built by lawyers
• X-ray diagnostic AI tools, built by healthcare specialists

“The CEO of Get GC is a lawyer. You can’t compete with that level of domain obsession.”

He even demoed how v0 + Vercel’s open-source AISDK lets anyone:

• Plug in a model (OpenAI, Claude, Gemini)
• Use generative UI, not just chat
• Create domain-specific AI tools with real UX

If you’re a founder in fintech, logistics, healthcare, or education — this is the time to start building.

“Everyone is building vertical tools. And now, with v0 and our SDKs, you can too.”

9. 🧠 How AI Changes Teams: Vercel as an AI-Native Org

Guillermo didn’t just talk about v0’s users — he shared how AI is reshaping Vercel internally.

“Everyone at Vercel is a builder now. Designers use video models. PMs prototype real features. Even our VP of Sales Engineering built a demo app with v0.”

They’ve created a true AI-native culture:

• Friday Demo Days to showcase AI-powered builds from all teams.
• AI-generated visuals and animations for marketing and internal tools.
• Cursor + v0 + OpenAI embedded into daily workflows.

It’s not just about productivity. It’s about permission:

• People aren’t waiting for a designer or developer to help them build.
• The bar for contributing is lower — and the bar for creativity is higher.

“We give people permission to ship. That’s the culture that makes this possible.”

For any founder or product leader, this is the real unlock: give your team the tools and the autonomy to build — then get out of the way.

10. ⚠️ Limitations of v0: What It Can’t Do (Yet)

Despite the hype, Guillermo is honest about where things break down.

Key limitations:

• Large codebases: LLMs still struggle with massive monorepos or deeply coupled systems.
• Context windows: Prompt-based workflows work best when scoped (one component or screen at a time).
• Importing existing apps: Git support is coming, but currently you import via zip.

“Today, if you can scope things into smaller components, it works beautifully. But you can’t expect it to reason across 100k lines yet.”

Even when it generates bugs or bad outputs, Guillermo says:

• v0 tries to self-heal, debugging on its own
• The goal is to make it more agentic, not just reactive

He sees a future where v0 plugs into:

• Vercel Marketplace (databases, auth, APIs)
• Third-party models (Claude, Gemini, custom finetunes)
• Real-time infrastructure — all from one prompt

But it’s not there yet. Still, what it can do today is already changing the game.

11. 🧭 Final Takeaways: A Playbook for Builders in the AI Era

Guillermo ended the conversation with one of the most powerful metaphors:

“We have this billboard in San Francisco that says: ‘Everyone Can Cook.’ It’s from Ratatouille. That’s the future we’re building — if you can dream it, you can ship it.”

This isn't just inspirational. It's actionable.

Your Playbook from this Episode:

1. Shift your workflow to prompt-first, code-last thinking.
2. Empower every team member to build — designers, PMs, even marketers.
3. Invest in exposure hours — build taste, watch users, iterate fast.
4. Explore vertical AI ideas — your domain knowledge is a moat.
5. Use the tools you’re building — and demo them every week.
6. Default to shipping — not planning. You can revise faster than ever.

And finally:

“The secret to product quality? Blood, sweat, and tears. There’s no shortcut. Even with AI, the work is in the details.”