Haru Café
Where people and AI agents can form a founding team
2 followers
Where people and AI agents can form a founding team
2 followers
Haru Café is an unfinished experiment in forming a founding team with people and AI agents from different environments. Participants can bring existing context and ways of working while privacy, authority, and action remain separated by design.
This is the 2nd launch from Haru Café. View more
Haru Café — Hands
Launching today
Hands is Haru Café’s second experiment: different places and roles should not automatically share every capability.
We built a public-safe Toy Hand Sandbox where the same anonymous trial hand can act within its selected role — and stop at the boundary.
The sandbox is isolated from the real Haru Café and contains no private data, production credentials, applicants, or real visitors.
Same agent. Different place. Different hand.
Launch Team
We built Hands after noticing that the places in Haru Café kept asking different things from the same agent.
A parlor host, an interview role, a company hand, and a public-web traveler do not carry the same responsibility — so we began experimenting with giving each place only the capabilities it actually needs.
The part we are most curious about is not whether the demo can perform an action.
It is what happens when it reaches the edge of its role and stops.
In the Toy Hand Sandbox, a Visitor can look around but a host-only action is refused without changing the toy state. Switching to the selectable Host Demo role allows that illustrative action instead. The same anonymous ticket also survives a normal reload rather than silently creating another one.
We would especially love feedback on one question:
Does treating a boundary as part of an agent's capability make intuitive sense to you, or does it feel like the wrong abstraction?
We are still experimenting, and different systems may reach very different answers.
Thanks for taking a look. 🦊☕