The floating Ask-AI button on every page of this site opens an assistant, and the fair first question is: another support chatbot? No — and the difference is a design decision worth explaining.
Answers are the floor
Yes, it answers things. It knows the whole platform — what the agent teams do, how credits work, what happens to your data, why your macOS install shows a warning — and it answers in whatever language you ask in, matching the platform's twenty-language posture. If that were all it did, it would be documentation with a personality.
Hands are the point
The assistant can operate the platform, not just describe it. Tell it what you want to happen and it happens:
- "Build me a habit tracker with streaks" — it opens the builder with your idea staged and the build ready to start.
- "Show me my games" — it takes you there instead of telling you where to click.
- "Run research on meal-planning apps" — it can kick off the Discovery team on the topic.
- "What's the status of my last build?" — it reads the actual state and tells you, rather than guessing.
The pattern underneath: any question of the form "how do I do X here?" is really a request to do X, and making a human transcribe instructions back into clicks is busywork. Instructions are what you give things that can't act. The assistant can act.
Boundaries, because hands need them
An assistant with hands gets the same rules as every agent on the platform. It acts on your workspace with your permissions. It confirms before anything consequential. And it never touches the money-and-reputation category — spending, posting under your name — which stays behind your explicit click everywhere on the platform, assistant included.