Rickard Hansson

Founder @ | Building Lightning-Fast MVPs in Public | Vibe Coding Advocate | 45-Minutes a Day Coding Series

[Rickard Hansson @rickardhansson]

I feel this. The token streaming bottleneck gets real when you're trying to iterate fast.

One thing that's helped: stop re-prompting for features that are already solved problems. Use pre-built components for stuff like chat or file sharing, let the AI focus on your actual product logic. Cuts way down on token waste and speeds up the whole cycle.

What features are eating most of your prompt budget?