Sovereignty · 2026-09-29 · 1:52

An agent with its own phone number

Name: An agent with its own phone number
Uploaded: 2026-09-29
Duration: 1 min 52 s
Description: Before you give an agent a name, a number and a wallet, ask three things: which payments and messages need your yes one by one, what it may never do with that number and address whoever asks, and where you see afterwards everything it said and paid in your name. If the answers are not on the page, t

First an AI test that measured access rather than intelligence, then an agent reading your bank feed. This week, the next step: Manus launched Cue on 28 September, where each agent "gets its own identity" — "its own email, phone number, wallet, and computer", able to send messages, pay within a budget you set, and take your calls.

In favour

Genuinely useful: an agent that answers the call you cannot take and pays the small bill without asking twice.
Agents can share a goal in a group chat and hand work to each other.

Worth watching

The change is from lending access to lending identity: the shop or the person on the line deals with the agent, not with you, while what it says and spends is yours.
The announcement describes the identity, not the controls. As RuntimeWire noted, it "does not explain in detail how users approve individual transactions, restrict what agents can do with their email or phone identity, or review actions after they occur." They may exist; we could not find them described.

Our takeBefore you give an agent a name, a number and a wallet, ask three things: which payments and messages need your yes one by one, what it may never do with that number and address whoever asks, and where you see afterwards everything it said and paid in your name. If the answers are not on the page, they are not yet in your hands.

Source: Manus, «Introducing Manus 2.0» (28 Σεπτεμβρίου 2026)

← All Focus posts

AI-narrated · AI-generated content

Transcript

First, an assistant that could read your documents. Then, one that reads your bank data. This week, an agent with its own identity. A new personal-agent app launched on 28 September. Each agent gets its own email, phone number, wallet and computer. It can pay within the budget you set, and take your calls. Notice what changed. Before, you lent an assistant your access. Now the agent becomes someone. The person on the call deals with it, not with you. Yet what it says and spends is still yours. The appeal is easy to see. An agent that answers the call you can't take, chases the reply you'd forget, and pays the small bill without asking you twice. Several agents can share a goal in a group chat, and hand work to each other. It's in early access, by invite code, on web, desktop and mobile. The announcement describes the identity. It doesn't describe the controls. One report noted it doesn't explain in detail how users approve individual transactions, restrict what agents can do with their email or phone identity, or review actions after they occur. Those controls may exist. We couldn't find them described. So before you give an agent a name, a number and a wallet, ask three things. Approval. Limits. Record. If the answers aren't on the page, they aren't in your hands yet. An agent that speaks for you needs what a new employee does: a clear mandate, a spending limit, and someone who reads the log.