Manus 2.0 has just given your AI assistant its own phone number, an inbox and a wallet — and that's a bigger deal than it sounds. These are the everyday credentials any human worker takes for granted. Putting them in an agent's hands changes what it can be trusted to do.
For the last couple of years, "AI agent" has mostly meant a chatbot in a browser tab. It could draft an email, but it couldn't send one from its own address. It could suggest a purchase, but it couldn't pay for it. Manus has now closed that gap: its agents come with their own phone numbers, email addresses, wallets and virtual computers, and a new app called Cue puts all of it in one place. Here's what actually shipped, and what it means for people who just want fewer tabs open.
What happened
Every agent gets its own identity
Think of it like hiring a remote assistant and handing them a desk, a phone line and a company card on day one. The desk is the virtual computer — a machine running in the cloud that the agent treats as its own, so it can open apps and click through screens without ever touching your laptop. The phone line and the inbox mean it can answer a call or send a message as a contact in its own right, instead of pretending to be you. And the wallet means it can actually pay for things.
That last part comes with a leash. Inside Cue, the agent only spends within a budget you set in advance. It's closer to a prepaid card you hand a teenager than to your own credit line.
Manus 2.0 starts work without being asked
Manus 2.0 also learned to fire off tasks when something happens — a new email lands, a message comes in, a calendar event hits. Instead of you typing "invoice this client" on the first of every month, the agent already knows that when that client emails, it should draft the invoice, send it and log it.
Agents can work as a team now too. Several of them split a project into stages and pass the results down the line, the way a small newsroom works: one reporter files, an editor tightens it up, someone checks the facts. India Today and Briefs both covered the rollout, and the AI channel @aioftheday logged 884 views and 26 reposts on its write-up — modest numbers, but the direction of travel is hard to miss.
What it means for you
Plain English: you're no longer buying a chatbot. You're renting something closer to a temp worker who has a phone, an address and a spending limit — and who never says "sorry, I'm off at five."
If you'd rather test the water before handing a bot your calendar and a payment method, there's a whole stack of free AI tools you can play with first. Same skills, zero risk.
At home and in everyday life
• The dentist's office calls back while you're in a meeting. Your agent picks up, grabs the earliest slot that's free on your calendar, and tells you about it afterwards.
• Groceries get reordered on a schedule, with the agent watching the total so it stays inside your budget cap.
• School and club emails get read, the dates get pulled out, and everything lands in your calendar. No more scrolling through a 60-message thread to find out when the trip is.
None of this is glamorous. That's the point. The value of an agent with its own phone number isn't that it's clever — it's that the boring middle of your day stops needing you in it.
At work, in business and for extra income
• Freelancers and consultants: point the agent at your inbox and let it reply to new inquiries with your rates and a booking link, then flag only the leads worth your own answer.
• Service businesses — salons, plumbers, tutors — get a phone line that never rolls to voicemail.
• There's a business in this, too. Setting up an "AI receptionist" for local shops is something you can sell: you configure the budget, the hours and the rules once, and charge a monthly fee.
• Small online shops: an agent that handles "where's my order?" at 2 a.m. while you sleep.
For study and creative projects
• Students: an agent that drops new readings into a folder the moment a professor posts them, with a short summary attached.
• Creators: two agents as a team, one researching and one drafting, with you as the editor. You get a rough first pass in minutes instead of staring at a blank page.
• Newsletter writers: an agent that watches for new posts and new messages and assembles a rough digest for you to shape.
The pattern is the same in all three cases. You stop being the person who copies things between apps.
How to try it right now
Free option first, and don't enter a card on day one.
1. Start free. Read the full breakdown of Manus 2.0 and Cue and write down three repetitive tasks you'd happily never do again.
2. Sign up for Manus and open Cue. Poke around with a boring job first — summarising an email thread, say — before you connect anything real.
3. Set the budget cap before you touch the wallet. Start with the smallest amount the app allows, the same way you'd start a new driver on a quiet street.
4. Give the agent a separate email address. Keep your main inbox clean and forward only what you want it to see.
5. Add one trigger. "When this sender emails, do X" is the simplest useful automation, and it teaches you how the whole thing behaves.
6. Only then let it spend. Watch the first five purchases closely. The cap limits the damage; it doesn't prevent a bad call.
7. Add a second agent and chain them. One researches, the other drafts. Watch for the moment a detail gets dropped in the handoff.
8. Check in daily for the first week. Agents drift. A two-minute review keeps them honest.
Upsides and what changes
The obvious win is time back. The more interesting shift is what an agent with its own identity can do that a chatbot can't: it can be reached. A supplier can email it. A customer can call it. It shows up in the world as a contact rather than as a feature buried in a website.
For solo operators, that's real leverage. A one-person consulting shop can now look and behave like a five-person firm on the outside — calls answered, emails acknowledged, invoices out on time — without hiring anybody.
For creators, it deletes the part of the job nobody signed up for: the chasing, the scheduling, the copy-paste.
And for the rest of us, it changes what delegation actually means. Until now, "delegating to AI" meant asking a question and then doing the work yourself. Now there's something you can hand a job, a budget and a deadline to — and it finishes on its own.
Limitations
An agent with a wallet is an agent that can be scammed. Phishing emails are already written to fool careful humans; a bot that acts on incoming mail without asking is a much softer target. The budget cap limits the damage, but it won't stop a wrong decision — and it definitely won't stop an embarrassing one, like a testy reply going out to your biggest client at 3 a.m. Agents that trigger off calendar events and incoming messages also act while you're not watching, so mistakes can pile up before anyone notices. Multi-agent handoffs are the newest and least proven part of the package: chaining three bots means more moving parts, and more moving parts means more ways for something to slip through. There are open questions too, like what happens to a phone number or an email address that isn't tied to a real person when a bank, a platform or a customer starts asking. The reporting confirms the features exist — nobody's promising it'll all go smoothly.
Conclusion
The direction is clear: AI assistants are getting identities, not just personalities. A phone number, an inbox, a wallet and a computer add up to something you can hand a job to, and the budget cap is what makes it safe enough to actually try. You don't have to hand over your business tomorrow. You do have to know this is on the table now, because the person you're competing with is already looking at it.
One action for today: open a notes app and list the three most repetitive tasks in your week. That's not busywork — that's the shortlist you'll hand to your first agent, and having it ready is the only thing standing between you and the free test drive.
Comments 0