No PriorsNo Priors

No Priors Ep. 92 | With StackBlitz CEO and Co-Founder Eric Simons

Sarah Guo and Eric Simons on bolt.new turns ideas into full-stack apps, redefining software creation.

Sarah GuohostEric Simonsguest
Dec 6, 202438mWatch on YouTube ↗
What Bolt.new is and how it differs from generic codegen toolsWebContainer technology and running full dev environments in the browserStrategic open-sourcing of prompts, code, and the role of communityNon-technical users and the emergence of “software composers”Bolt as a real-world benchmark for new codegen modelsReal startup and product use cases built entirely with BoltWhy this generation of AI codegen surpasses traditional no-code platforms

In this episode of No Priors, featuring Sarah Guo and Eric Simons, No Priors Ep. 92 | With StackBlitz CEO and Co-Founder Eric Simons explores bolt.new turns ideas into full-stack apps, redefining software creation StackBlitz CEO Eric Simons discusses Bolt.new, an AI-powered browser-based tool that generates full-stack, production-grade web applications from natural language prompts. Built on StackBlitz’s WebContainer technology, Bolt runs dev environments entirely in the browser, avoiding backend setup, latency, and cloud costs. Simons explains why they open-sourced Bolt’s prompts and code, how community usage drives both product learning and model evaluation, and why this moment is fundamentally different from earlier no-code attempts. He also highlights real users launching profitable startups with massive cost and time savings, and predicts a rapid shift toward “software composers” directing powerful codegen systems.

At a glance

WHAT IT’S REALLY ABOUT

Bolt.new turns ideas into full-stack apps, redefining software creation

  1. StackBlitz CEO Eric Simons discusses Bolt.new, an AI-powered browser-based tool that generates full-stack, production-grade web applications from natural language prompts. Built on StackBlitz’s WebContainer technology, Bolt runs dev environments entirely in the browser, avoiding backend setup, latency, and cloud costs. Simons explains why they open-sourced Bolt’s prompts and code, how community usage drives both product learning and model evaluation, and why this moment is fundamentally different from earlier no-code attempts. He also highlights real users launching profitable startups with massive cost and time savings, and predicts a rapid shift toward “software composers” directing powerful codegen systems.

IDEAS WORTH REMEMBERING

7 ideas

AI codegen has crossed a tipping point for real production apps.

Simons argues models like Claude 3.5 Sonnet now reliably generate accurate, production-grade code, enabling end-to-end apps rather than just snippets or demos—something that wasn’t viable even earlier this year.

Running dev environments in the browser unlocks speed and scalability.

StackBlitz’s WebContainer OS executes full toolchains (Next.js, Vite, npm installs, etc.) client-side, eliminating server latency and per-minute cloud costs while allowing arbitrary packages and realistic app stacks.

Open-sourcing prompts and glue code builds durable advantage via ecosystem.

Bolt’s team believes their moat is end-to-end product quality and growth, not secret prompts, so they open-sourced them to spur contributions, credibility, and widespread adoption rather than behaving like a fragile GPT-wrapper.

Community is now essential to teaching people how to use AI tools.

Because AI outputs are non-deterministic and prompt quality matters, StackBlitz leans on power users to share workflows, tutorials, and best practices—reducing churn and making users often more expert than the creators themselves.

Non-technical professionals can now realistically build and launch products.

Entrepreneurs and PMs who understand product requirements but not code are using Bolt to ship full apps—often replacing $5K–$30K dev quotes with $50–$200 subscriptions and compressing timelines from months to weeks.

Bolt is emerging as a de facto benchmark for new codegen models.

Researchers and practitioners plug open-source models into Bolt Local to see if they can power realistic, complex app-building workflows—creating a practical “can it run Bolt?” test beyond academic coding benchmarks.

The future developer role shifts from coder to high-level software composer.

Simons predicts developers will increasingly direct AI agents with higher-level instructions, focusing on system design and hard problems while AI handles boilerplate, UI generation, and routine coding tasks.

WORDS WORTH SAVING

5 quotes

Bolt is kind of similar to ChatGPT or Claude, except you use it to build full-stack web applications.

Eric Simons

We’re not gonna win because of our system prompts. We’re gonna win by growing extremely quickly and building the best end-to-end product experience.

Eric Simons

It turns out, managing an AI is extremely similar to managing actual software developers.

Eric Simons

This is the most incredible arbitrage opportunity in web development ever.

Eric Simons (quoting a user tweet)

AI code gen models have gone over the tipping point of being good enough to really write real applications that are production grade.

Eric Simons

QUESTIONS ANSWERED IN THIS EPISODE

5 questions

How will traditional web development agencies and freelancers adapt their business models as tools like Bolt compress cost and delivery times so dramatically?

StackBlitz CEO Eric Simons discusses Bolt.new, an AI-powered browser-based tool that generates full-stack, production-grade web applications from natural language prompts. Built on StackBlitz’s WebContainer technology, Bolt runs dev environments entirely in the browser, avoiding backend setup, latency, and cloud costs. Simons explains why they open-sourced Bolt’s prompts and code, how community usage drives both product learning and model evaluation, and why this moment is fundamentally different from earlier no-code attempts. He also highlights real users launching profitable startups with massive cost and time savings, and predicts a rapid shift toward “software composers” directing powerful codegen systems.

What safeguards or best practices are needed to ensure security and maintainability of AI-generated production code at scale?

How might the definition of a “software developer” change as more non-technical users become effective software composers through tools like Bolt?

Could the heavy optimization of models for code generation create blind spots or trade-offs in other capabilities, and how should labs balance that?

What new types of products or business models become possible when the marginal cost of building a full-stack app approaches zero?

EVERY SPOKEN WORD

Install uListen for AI-powered chat & search across the full episode — Get Full Transcript

Get more out of YouTube videos.

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

Add to Chrome