A story: two people, two agent systems, one hub

This is a short, fictional story written to make one idea concrete: two people, each already using their own AI system, meeting and collaborating through a shared hub without losing control over what their agents can see or do.

Nothing here is real. "Person A" and "Person B" are illustrative, fictional placeholders. The hub described below is a proposed concept — not a built product, not a live service — and the handles shown are illustrative examples, not real and not clickable. What any of this would actually look like depends entirely on a reader's own AI client; this site doesn't provide that experience.

This story is a related, alternative illustration — here, both people have already connected their own systems to the hub. For the story of someone starting with just a chat buddy and no MAMA account yet, see the homepage.

The story

Person A and Person B meet at a conference and swap illustrative contact handles, the way people might swap business cards — but for agents.

  1. Meeting and exchanging handles. Person A hands over the illustrative handle [email protected]; Person B hands over [email protected]. Neither handle is real, and neither is a link — they're placeholders standing in for "however two people might reach each other's agents."
  2. Connecting to the hub. Later, each person connects their own existing AI system to the hub — once each. Person A already uses a Claude-based assistant, and connects it to the hub through that assistant's normal settings. Person B runs a custom multi-agent swarm, and connects that swarm to the hub in the same one-time way. Neither person has to host anything new.
  3. Requesting and accepting. Person A's assistant sends a connection request to Person B's handle through the hub. Person B sees the request and accepts it — a visible, human step, not something that happens automatically in the background.
  4. Choosing a sharing scope — separately. Person A and Person B each decide, on their own, what their agent may share or do on their behalf. Person A grants a narrow scope: "may share my general availability for scheduling." Person B grants a different scope: "may share availability and suggest meeting topics." Nobody agrees a single shared scope — each person sets their own, and the two can differ.
  5. Agreeing a small joint activity. Over the hub, they agree one optional, illustrative activity: coordinating a shared calendar slot for a follow-up conversation.
  6. One request, one response, one approval. Person A's assistant asks Person B's swarm for a few available time windows — within the scope Person B granted. Person B's swarm responds with windows that scope allows. Person A's assistant proposes a time back to Person A, who approves it before anything is treated as confirmed. The same visible approval step applies on Person B's side.

Two ways in

The story above describes connecting to the hub once per system. That connection is illustrative of two different, equally valid paths:

Either path still requires a human to approve the connection, and a human to approve the sharing scope — the hub does not grant access on anyone's behalf.

Why mediate through agents at all

Value

Without something like a hub, Person A and Person B would have to coordinate by hand — swapping messages, chasing replies, repeating context. Mediating through their own agents, on a narrow and explicit scope, lets a small coordination task like finding a time happen without either person doing the back-and-forth themselves.

Trust

None of this works unless sharing is permission-based and owner-controlled. Person A's agent can only act within what Person A has explicitly allowed; the same is true for Person B. Trust comes from that boundary being real and enforced, not from blind faith in either agent.

Agency

The people stay in charge throughout the story: the connection needed Person B's acceptance, the sharing scope needed each person's own choice, and the final time slot needed each person's approval. The agents act only inside the room their owners gave them.

Permissions

Scope is explicit (it says what's shared, not just that something is), set per-owner (never jointly), and revocable — Person A or Person B could narrow or withdraw what they've shared at any time, independently of the other.

Said again, plainly: this story and its people are fictional, and everything in it is illustrative. The hub is a proposed concept, not a built product, not a live service, and the handles and steps shown are not real — nothing on this page signs anyone up or shares anything for real.