Your next AI assistant won’t just live in a chat window — it’ll have its own phone number, its own email address, and its own wallet. Manus launched Manus 2.0 on Monday alongside Cue, a standalone app for personal AI agents, and the pitch is a genuine leap: agents that hold accounts, spend money, and take calls on your behalf.
According to Bloomberg, each Cue agent gets its own email address, phone number, digital wallet, and computer. Agents can send messages, make payments within a budget you set, take calls and leave summaries, and hand work to each other in group chats. One agent researches venues, another shortlists options, a third drafts the presentation — coordinating among themselves while you make the final call.
From Chatbot to Account-Holder
This is the permission-model shift the whole agent industry has been circling. A chatbot answers questions; an account-holder acts in the world. Manus’s examples make the distinction concrete: an agent that scans a restaurant QR code to place your order or hold your spot in line, one that takes a call while you’re busy and leaves a summary, one that keeps working after you’ve closed the app.
Manus 2.0 wraps the concept in infrastructure: a Cloud Computer environment for tasks that need an always-on machine, Automations that trigger on schedules or events from connected email, calendars, Slack, or Notion, and an expanded Manus Studio spanning documents, spreadsheets, PDFs, slides, websites, code, games, and video. The company claims its new Cascade harness uses 23.2% fewer tokens, 28.2% less time, and 32% less cost in one tested configuration — figures that are, for now, unverified.
Cue is in early access on web and desktop (free with an invitation code), with mobile coming later.
The Breakup That Made This Possible
The launch lands weeks after one of the year’s strangest corporate sagas. Meta agreed to acquire Manus for $2 billion — then Beijing ordered the deal canceled in April, forcing the Singapore-based startup back to independence. Manus is now reportedly in talks to raise $500 million at a $4 billion valuation: roughly twice what Meta was willing to pay.
The timing also matters competitively. September has been agent-launch month across the industry: Meta’s own “Muse” agent shot to the top of app store charts, while Google, Apple, and Microsoft all shipped agent products. As one AI industry tracker notes, Manus is betting that identity — agents with persistent accounts and resources — is the differentiator in a suddenly crowded field.
The Trust Question
An agent with a wallet is also an agent that can be fooled into spending. Just last week, Salt Labs disclosed an email-borne prompt-injection flaw that let researchers run code inside a victim’s Manus environment (since patched through Meta’s bug-bounty program, back when the two companies were still entangled). Manus’s answer is user-set budgets and permission controls — but the security model for account-holding agents is being written in real time, and Cue is now its biggest live test.
Why this matters
The commodity coverage will treat Cue as another agent launch. The real story is structural: Manus went from acquisition target to independent rival in five months, doubled its would-be price tag, and shipped the most aggressive version of the agent thesis — not a smarter chatbot, but a digital employee with credentials. If Cue’s permission model holds up under real-world prompt-injection attacks, every other agent maker will have to follow it into the account-holder era. If it doesn’t, the wallet becomes the cautionary tale. Either way, the industry just got its test case.
Meanwhile in Apps & Tools: how talking AI avatars work — and which ones you can actually use. And Google Pics: what it is and how to use it.
