TL;DR
In this episode of The Good Thing, Stefan and Jens unpack the fallout from Cursor’s sudden pricing changes, discuss Microsoft’s positioning with GitHub Copilot, and explore how AI tools affect developer productivity. The conversation highlights the tension between hype, distribution, and building real products in developer tooling.
Cursor Pricing Backlash and Plan Changes
After covering industry updates like PlanetScale and the news of Lee Robinson leaving Vercel, Stefan and Jens shift to Cursor’s abrupt pricing adjustments. What was marketed as “unlimited” became capped at 500 requests, then reduced again to around 200. Users shared frustration in online forums, seeing the changes as poorly timed and confusing.
They just kind of changed their pricing and messed a lot of things up
Jens adds that repeated shifts can signal deeper problems inside a company under pressure.
Cracks in the Business Model
Cursor’s situation raises broader questions about AI developer tools. Jens reflects on conversations with investors, pointing out that raising capital without clear margins is risky if costs stay high.
There’s a lot of evidence that behind the scenes something is going really badly
Stefan speculates that Cursor has raised large sums but hasn’t built its own models yet, relying instead on wrappers around existing ones.
GitHub, VS Code, and Microsoft’s AI Advantage
The conversation turns to Microsoft. With GitHub and VS Code, they already control distribution. Adding Anthropic’s Claude and OpenAI’s models to Copilot puts enormous pressure on smaller AI startups.
If you buy a Microsoft license for 100 bucks and for ten more you get AI capabilities, that’s it. The CFO won’t approve another 50-dollar AI tool.
For the hosts, Microsoft’s patience and infrastructure give it a decisive edge.
Do AI Coding Tools Really Improve Developer Productivity?
The episode also explores whether AI makes developers faster. Claims like “50% of code is written by AI” sound impressive, but Jens warns that volume doesn’t equal quality.
Every line of code is a liability. More lines don’t mean better results.
Stefan agrees, noting that tools like Cursor are great for small projects but unreliable in large, complex codebases.
The Bigger Problem: Collaboration
Toward the end, the hosts connect this debate to their own work. They argue that coding tools only speed up once the specification is already clear. The harder challenge lies in collaboration across product, design, frontend, and backend teams.
The real bottleneck is better collaboration.
They frame this as the real frontier of developer experience.
Closing Thoughts
The episode closes with a reminder: AI tools can accelerate workflows, but business models, developer trust, and team coordination remain the real tests. Cursor’s struggles are part of a bigger story about how developer platforms will evolve under pressure from incumbents like Microsoft.
This episode was directed by Jacob Javor. Transcript lightly edited for clarity and flow.
