Skip to main content
The Rework Podcast

Building with LLMs, Finding a Co-founder & Other Listener Questions

25 min episode · 2 min read

Episode

25 min

Read time

2 min

Topics

Relationships, Startups, Leadership

AI-Generated Summary

Key Takeaways

  • LLM Development Limits: LLMs enable designers to prototype quickly without programmers, but production code at scale requires human review due to security risks. Use LLMs for non-critical apps or prototypes, then bring experienced developers for data-sensitive applications.
  • Cofounder Discovery Strategy: Meet potential cofounders through extended exposure in college programs or open source communities rather than speed-dating approaches. Work together first to test compatibility before formalizing partnerships, similar to dating before marriage, avoiding contractual commitments with strangers.
  • Founder Exit Planning: Small companies with 40-80 employees rarely transition successfully to non-founder leadership. Founders who remain involved as passive owners create dysfunction. Clean separation through sale provides better outcomes than attempting generational handoffs in software businesses.
  • Product Idea Generation: Develop ideas through direct exposure to business problems while working for others, not isolated brainstorming. Internal validation through personal use matters more than external surveys. Basecamp remains 37signals' most successful idea after 22 years despite launching multiple products.

What It Covers

Jason Fried and David Heinemeier Hansson answer listener questions about using LLMs for development work, finding cofounders, succession planning for founder-led businesses, and generating product ideas through real-world exposure.

Key Questions Answered

  • LLM Development Limits: LLMs enable designers to prototype quickly without programmers, but production code at scale requires human review due to security risks. Use LLMs for non-critical apps or prototypes, then bring experienced developers for data-sensitive applications.
  • Cofounder Discovery Strategy: Meet potential cofounders through extended exposure in college programs or open source communities rather than speed-dating approaches. Work together first to test compatibility before formalizing partnerships, similar to dating before marriage, avoiding contractual commitments with strangers.
  • Founder Exit Planning: Small companies with 40-80 employees rarely transition successfully to non-founder leadership. Founders who remain involved as passive owners create dysfunction. Clean separation through sale provides better outcomes than attempting generational handoffs in software businesses.
  • Product Idea Generation: Develop ideas through direct exposure to business problems while working for others, not isolated brainstorming. Internal validation through personal use matters more than external surveys. Basecamp remains 37signals' most successful idea after 22 years despite launching multiple products.

Notable Moment

David admits he cannot sit in the passenger seat of any business he owns, preferring to drive it into a wall rather than watch someone else steer, revealing why founder transitions fail even at billion-dollar companies like Starbucks and Dell.

Know someone who'd find this useful?

Episode Transcript

Welcome to Rework, a podcast by thirty seven Signals about the better way to work and run your business. I'm Kimberly Rhodes with Jason Fried and David Heinemeier Hanssen, cofounders of thirty seven Signals. This week, we are knocking out some listener questions. We got some video questions in, which we love to receive. The first one is from Ryan. This is Ryan Chitwood. I'm an independent iOS app developer on the side. And since I work by myself, I've been wondering about adapting your setup of one designer and one developer comprising a team. What do you think about these days with the advent of LN tools like Cloud Code? You know, these are getting pretty good at coding at this point. What do you think about teams of one, that being a designer with some development skills, using a Cloud Code or something like that as filling in the developer portion there. I don't know if y'all are ever thinking about moving to that model at 37 signals, but what do you think about for an independent person like me? Thanks. David, I'm gonna start with you. And then, Jason, I'm sure you have some things to add on the designer side. I think this is one of the most amazing things about the LLMs, and it's something we've seen already at 37 signals, in a big way. JC, in particular, have been really good on our team of just running ahead of programmers, and he is able to produce so much more, so much faster with the LMS than he was before. Now it all depends on what kind of fidelity you need. At our scale, the number of customers we have and the criticality that we're dealing with, we can't usually ship a bunch of LOM code into production. It's still just not good enough for that yet. There's obvious errors, obvious mistakes. There's barely a day that goes by right now without someone getting exposed online for shipping some LLM slop that then left a huge security hole. There was just this whole bit about tea. I don't know if you saw this this app where Oh, that that tea app. Women could talk about their experiences with men. And, apparently, it was done by this guy who literally had six months of experience as a programmer on a boot camp, and I'm sure used a bunch of LLMs, and everything was basically just wide open. So here you had a runaway success app that spilled the tea on 2,000,000 women because it basically wasn't locked down. And that's, of course, sort of the nightmare scenario. Not that these things can happen with, quote, unquote, real programmers, human programs, whatever you wanna call it, but it's far less likely that it's going to happen. So I would look at what kind of apps are you working on? Are you working on things where if use of data got leaked, it's potentially a major, major problem? …

Get the full transcript (5,493 words) + summary by email — free

One-time email with the complete transcript and AI summary of this episode. No account needed.

One email, no spam. We’ll also show you what SignalCast does.

Browse all The Rework Podcast transcripts →

You just read a 3-minute summary of a 22-minute episode.

Get The Rework Podcast summarized like this every Monday — plus up to 2 more podcasts, free.

Pick Your Podcasts — Free

Keep Reading

Books, tools, and gear mentioned in this episode

SignalCast may earn commission on purchases via these links.

company

  • Basecamp remains 37signals' most successful idea after 22 years despite launching multiple products.
  • Basecamp remains 37signals' most successful idea after 22 years despite launching multiple products.

More from The Rework Podcast

We summarize every new episode. Want them in your inbox?

Similar Episodes

Related episodes from other podcasts

Explore Related Topics

This podcast is featured in Best Business Podcasts (2026) — ranked and reviewed with AI summaries.

Read this week's Startups & Product Podcast Insights — cross-podcast analysis updated weekly.

You're clearly into The Rework Podcast.

Every Monday, we deliver AI summaries of the latest episodes from The Rework Podcast and 192+ other podcasts. Free for one show.

Start My Monday Digest

No credit card · Unsubscribe anytime