How Veto was built?
Validating user flows by letting AI generate clickable prototypes from a flow chart
I hand the flow chart to AI and get a working hi-fi back, collapsing the gap between an idea and something stakeholders can click. A flow chart is abstract; a clickable prototype is a conversation. People could react to the real interaction in hours, not weeks, not to a static picture of it.
Try it at:
What it Veto: an AI that triages your inbox, then waits for you to approve, fix, or kill each move.
Why Veto: Veto takes over the repetitive inbox work, triaging, drafting, filing, so a full inbox goes from an hour of doing to a few minutes of deciding.
Reflection
Finding the balance between human and AI control
Building Veto got me to the question I actually care about: not how much an agent can do, but where it should act on its own and where the person stays in charge. The pattern I landed on, propose, explain, let the human override, isn't specific to email; it's how I'd approach any agent that acts on your behalf. That balance is what I want to keep designing for.