Steve Jobs Vision Prover Connector for AI agents.

It forces your AI to act as a rigorous design critic, ensuring your product is simple and focused. It pushes the AI to think about what to remove rather than just what to add.

Yes, it requires the AI to explicitly name what it's removing and why that makes the product better. It prevents the AI from just listing features without thought.

It is a design failure where the system pushes hard decisions onto the user via settings menus. The Connector forces the AI to make those decisions automatically for a smoother experience.

It works for any product, including hardware, by focusing on the human experience and ownership. It ensures the core experience is owned end-to-end.

It flags fragmentation if your product relies too much on external APIs for its core value. It forces the AI to own the experience rather than renting it from others.

It will reject ideas that are bloated, tech-first, or lack clear design taste. It provides coaching to help you fix those specific issues.

No. It computes nothing and generates nothing. The LLM designs the product. this capability validates that the reasoning behind the design is rigorous. It catches contradictions: if the LLM claims zero configuration but describes a settings panel, the capability rejects and explains why.

A prompt says 'think like Steve Jobs.' The LLM nods and generates bloated designs anyway. This capability forces the LLM to fill in specific fields. name what it killed, describe the human moment without tech words, explain how the system decides. Capability calls are obligations. Instructions are suggestions. The LLM cannot skip the reflection.

Yes. The principles apply universally. A CLI capability with 47 flags is the same failure as a consumer app with a settings menu. you pushed decisions to the user. The 'experience backwards' pivot works for developers too: start with the developer's workflow, not your architecture.