The REWORK Podcast — selected coding-agent conversations
AI Revisited – Part 2
Seven episodes on agent-assisted development, durable software, product judgment, and agent-ready interfaces
A manually reviewed selection from The REWORK Podcast following 37signals’ move from cautious LLM assistance to terminal agents, AI-accelerated Basecamp development, agent-accessible product interfaces, and agent-built competition. Jason Fried and David Heinemeier Hansson emphasize disposable prototyping, senior review, architecture, product scope, and consequence-aware use. Their examples are first-party experience and forecasts, not controlled measurements.
In this collection Browse 7 summaries 5 of 7
REWORK host Kimberly Rhodes speaks with 37signals co-founder and CEO Jason Fried about using Claude in design work and deciding how AI should enter 37signals products. His examples focus on interactive prototyping and reversible local data rather than treating generated code as a finished implementation.
Key Points Covered
- Prototype without waiting on a busy teammate: Fried uses Claude to implement a recent-chat interaction so he can experience the idea while the person he had already asked is occupied with other work 03:38–06:35(opens in a new tab).
- Generate realistic local test data: Claude populates a local Basecamp database with multi-user conversations, attachments, and varied messages, then removes the temporary data after the design exercise 08:18–11:09(opens in a new tab).
- Separate native features from external agents: Fried expects selective embedded AI alongside customers bringing agents into Basecamp through simpler surfaces and command-line access 13:56–20:01(opens in a new tab).
- Keep optional power out of the way: AI should help interested customers without making the product harder for people who do not want it 21:05–24:56(opens in a new tab).
Full video: https://www.youtube.com/watch?v=EwFT3h55oOc(opens in a new tab)