Meta has acquired Manus, a Singapore-based startup behind a general-purpose autonomous agent it plans to fold into its products. Meta Manus AI agents are being framed as the next step beyond chat, software that can carry work through to completion.
Meta and Manus say the agent can independently handle tasks like market research, coding, and data analysis. The expectation is, you keep working where you already are, and the agent takes on more of the busywork that normally eats a day.
If you already use Manus, the companies are stressing continuity. The subscription service will keep running through the current app and website, and Manus will continue operating from Singapore.
Manus is built to execute
Manus is described as an execution layer that can take a goal, move through the steps, and deliver an output without constant back-and-forth.
To back that up, Manus points to scale metrics. It says the agent has processed more than 147 trillion tokens and powered the creation of over 80 million virtual computers in a few months. Meta says it will keep operating the Manus service while it works on integrations across its consumer and business products, including Meta AI.
Why Meta wants agents everywhere
Meta is positioning the acquisition as an accelerator for business AI, and a way to bake automation into products people already use. That matters because it suggests agents will not stay locked inside a standalone tool, Meta says it wants them inside its broader ecosystem.
The deal terms are still thin in public. CNBC reports the Wall Street Journal pegged the acquisition at more than $2 billion, and CNBC also reports Manus claimed an annualized average revenue above $100 million eight months after launch, with a run rate over $125 million. Those figures help explain why Meta would buy, not just partner.
What businesses should watch next
The near term looks like a two-track plan, keep the existing subscription steady while Meta starts weaving the agent into more surfaces. The practical questions for businesses are where it appears first and what data and tools it can access.
If you run a team, start by listing the workflows you would actually trust an agent to execute end to end, like recurring research, first-pass analysis, or internal coding tasks. Then watch for updates around admin controls, pricing tiers, and data handling language as Meta scales Manus to more businesses.