Meta Acquires AI Agent Startup Manus in Deal Valued Over $2 Billion
Meta Platforms Inc. has moved to acquire AI agent startup Manus in a transaction reportedly valued between $2 billion and $3 billion, according to information reviewed by toolmesh.ai. The deal marks a significant expansion of Meta’s artificial intelligence strategy, shifting focus from generative models that produce text to autonomous agents capable of executing complex workflows.
Strategic Pivot to Autonomous Agents
The acquisition stands as Meta’s third-largest to date, trailing only the purchases of WhatsApp and Instagram. Under the terms of the agreement, Manus co-founder and CEO Xiao Hong is expected to join Meta as a Vice President, reporting directly to Chief Operating Officer Javier Olivan.
While Meta has invested heavily in recruiting researchers to build large language models (LLMs), the Manus acquisition addresses a gap in consumer-grade AI applications. The move signals a transition from developing "smarter brains"—systems that answer questions—to deploying "hands and feet"—agents that perform actions. This distinction is central to the emerging "agent war" in Silicon Valley, where the utility of AI is measured by its ability to deliver finished results rather than preliminary information.
From Browser Kernels to Cloud Virtualization
Manus differentiates its technology by bypassing the limitations of standard chatbots. While a chatbot might generate a plan, an agent is designed to execute it. To achieve this, the startup initially spent over six months attempting to build an AI-native browser by modifying the open-source Chrome kernel. The team eventually discarded this approach, determining that browsers were too restrictive as containers for autonomous work.
Instead, Manus adopted a "cloud computer" architecture. By equipping the AI with a dedicated virtual machine, the system can interact with digital environments much like a human user. This setup allows the agent to execute command-line operations, such as git clone, download open-source projects, and install software within its own virtual environment.
The operational capacity of this architecture was illustrated during an internal test described by the company. When an agent was tasked with checking a train schedule on a website that was down for maintenance, it did not report an error. Instead, the system autonomously searched for contact information, drafted an inquiry email, and attempted to register a new email account to send the message.
Redefining Software Development
The integration of autonomous agents is expected to disrupt traditional software development models. According to materials reviewed by toolmesh.ai, Manus reports that approximately 80% of its internal code is now generated by AI. This trajectory suggests a shift where human engineers evolve into architects and reviewers, managing multiple coding agents simultaneously rather than writing syntax manually.
This development points toward a "no software" future where the speed of iteration accelerates dramatically. The company posits that the Software-as-a-Service (SaaS) sector faces significant disruption, as users may eventually rely on agents to build bespoke solutions for non-standard needs on demand, reducing the reliance on pre-packaged applications.