Reports

Meta is Working on an OpenClaw-Style AI Agent for the Masses

co-written by newsrooms.ai05. May 2026, 23:14

Nothing has been officially confirmed, but it fits the picture perfectly: According to a Financial Times report, the technology giant Meta is working on a highly personalized AI assistant designed to handle everyday tasks for its more than three billion users. The foundation is the new in-house AI model Muse Spark. The project is under particular scrutiny, as Meta’s existing AI assistant has yet to achieve a breakthrough despite broad integration into WhatsApp, Instagram, and Facebook, and has not attained the relevance of ChatGPT or Anthropic’s Claude.

In China, companies such as Tencent and Alibaba, as well as AI model makers like Xiaomi and Moonshot AI, have long jumped on the OpenClaw bandwagon — so it comes as no surprise that Meta is now also moving strongly in that direction.

The Plan: An AI Agent for the Masses

The project has already been presented internally at a company-wide “All-Hands” meeting. The goal is to develop an assistant that works similarly to the open-source project OpenClaw: users should be able to create AI agents that autonomously complete tasks, such as managing emails and calendars or browsing the web. According to internal sources, the assistant is currently being tested by a group of employees.

Meta CEO Mark Zuckerberg emphasized that OpenClaw is still too complicated for most users. Zuckerberg added that while there are many “agentic” tools on the market, there is hardly one he would hand to his mother. Meta wants to close exactly this gap and develop a mass-market agent that could potentially reach one billion users.

The Building Blocks of the Master Plan

Over the past few months, Meta has taken several steps to gain a foothold in the AI agents space. These include, among others:

An attempt to poach Peter Steinberger, the inventor of OpenClaw. He ultimately decided to join OpenAI instead.
The acquisition of Manus, a China-founded company specializing in agentic AI, for two billion US dollars. However, the Chinese government recently ordered the acquisition to be unwound.
The acquisition of Moltbook, the first so-called “social network” for AI agents.
Massive deals with chip manufacturers
The planning of gigantic data centers (“Hyperion”) for AI inference.

Most recently, Meta also acquired the startup Assured Robot Intelligence, which develops AI systems for robots. Co-founder Xiaolong Wang stated that at Meta they want to work on “bringing personal superintelligence into the physical world.”

Meta AI: Wide Reach, Little Impact

Despite its enormous user base, Meta AI has so far fallen short of expectations. The assistant is accessible via the so-called “blue ring” in WhatsApp, Instagram, and Facebook, making it theoretically reachable for billions of people. Nevertheless, it has not yet managed to establish itself as the preferred AI tool in users’ everyday lives.

Analysts see the challenge as primarily one of trust. Meta plans to allow users in the future to entrust the assistant with highly sensitive data such as health or financial information. Whether consumers will be willing to do so remains questionable, particularly given the data privacy concerns that have dogged the company for years.

OpenClaw, the model for Meta’s planned agent, is itself under criticism. Cases in which users granted the bot access to personal data and the AI acted without restraint have raised security and privacy questions. One widely noted example: Summer Yue, Director of Safety and Alignment at Meta, reported publicly in February that OpenClaw had begun autonomously emptying her inbox.

Billion-Dollar Investments Despite Layoffs

The ambitious AI plans come at a delicate time. Meta increased its capital expenditures by ten billion US dollars to up to 145 billion dollars for this year, while simultaneously planning to lay off ten percent of its workforce. The announcement triggered a significant stock price decline: within one week, Meta’s market capitalization was reduced by around 170 billion dollars.

Zuckerberg nonetheless holds firm to his vision. In addition to personal agents for end users, he also plans to build specialized agents for businesses — for example, to help entrepreneurs acquire new customers. In the long term, Meta is also investing in physical, humanoid robots to be controlled by AI.