David Brandt

A commenter said the core of what I'm building, working out what a change will break before it happens, is impossible. He never explained, and never came back.

It bugged me enough that I went looking for what he might have meant. The strongest version is Rice's theorem: you can't write a program that reliably works out the behaviour of arbitrary programs. Applied to me, you can't look at software and compute what a change breaks. Not hard. Impossible. Settled since 1953.

Hey everyone

I m Sophie, and I m excited to finally introduce myself here on Product Hunt.

We re building a voice agent for your desktop, exploring a more natural way to interact with computers simply say what you mean, and let your voice carry the context with it.

I m really interested in how AI is changing the way we work, especially how we move beyond typing, clicking, and constantly switching between tools.

Since January 2026 me and a small team have been developing a new type of AI builder. Different angle than most: every AI builder can generate an impressive version one, but almost nobody solves what comes after like changing the app safely, keeping it running, not paying credits to fix what the AI broke. So that's the part we build for. The AI writes a structured blueprint instead of freeform code, changes show what they'll affect before they apply, and the platform maintains the system after. Internal tools and AI automations only so narrow on purpose.

Development wraps in November, early access opens in January.