Meta Deploys AI Agents on Telegram via Manus AI
Meta is embedding AI agents into messaging platforms, beginning with Telegram. The agents, which are described as being similar to OpenClaw, were launched by Manus AI, a company Meta recently acquired. The move signals a push to compete with Telegram's existing agent ecosystem, with a WhatsApp integration expected to follow.
- The acquisition of Manus AI, a Singapore-based startup with Chinese roots, cost Meta over $2 billion, making it one of the company's largest deals. Manus founder Xiao Hong is now a VP at Meta, though the company will continue to operate its services independently. - The Manus agent on Telegram is not a simple chatbot; it is the full reasoning engine capable of multi-step task execution. Users can choose between the Manus 1.6 Max model for complex creative and reasoning tasks, or Manus 1.6 Lite, which is optimized for speed on lightweight requests. - Unlike OpenClaw, which is an open-source agent that runs locally on a user's machine, the Manus integration is designed for ease of use, connecting via a QR scan without requiring command line setup or API token management. This approach avoids some of the security risks associated with self-hosted agents, which can execute shell commands and have been flagged for potential vulnerabilities like leaked API keys. - The agent's capabilities go beyond chat to include generating structured documents like PDFs and reports, creating websites and slide decks, transcribing voice notes, and analyzing uploaded files. It also features long-term memory to recall user preferences and can integrate with third-party tools like Gmail, Calendar, and Notion. - Meta's broader strategy, articulated by Head of Business AI Clara Shih, is to provide every business with an AI agent, making it as fundamental as a website for customer interaction. The focus is on enabling small businesses to automate customer engagement without needing dedicated AI engineering teams. - Telegram already hosts a significant ecosystem of AI agents, particularly within the Web3 space, where bots are used for everything from customer support to automating user onboarding. Some projects in this space have reported user engagement rates as high as 25-35% with their AI agents.
