Meta's Chinese AI startup launches Manus Agents, emphasizing low barriers to compete with OpenClaw

As open-source project OpenClaw flourishes, founder Peter Steinberger has joined OpenAI. Recently, Meta’s Chinese AI startup launched Manus Agents, which also focus on personal agent services that can be operated directly within messaging apps. The product’s initial integration is with Telegram, emphasizing no need to set up servers, tokens, or configuration files—just scan a QR code to get started. It is open to all subscription tiers and will expand to more chat platforms in the future, directly competing with open-source counterparts like OpenClaw.

(The father of OpenClaw, Peter Steinberger, joins OpenAI! Altmam is optimistic about agent collaboration)

Agents Entering Chat Scenarios, Lowering Usage Barriers Is Key

Manus states that while the concept of “AI Agents in chat” has existed for some time, most solutions require command-line operations, API configurations, or ongoing maintenance, making it difficult for average users to use long-term. The core design goal of Manus Agents is to enable users to activate full agent capabilities with a single message within existing communication scenarios, without worrying about underlying deployment and operation.

Users only need to open the Agents page in the Manus workspace, scan the QR code, and connect to Telegram in about a minute. The entire process requires no server setup, tokens, or configuration files, significantly lowering the barrier for personal agent use.

Full Agent Capabilities, Not Just Chat Plugins

The official emphasizes that Manus Agents are not simple chatbots but an extension of the full Manus system within dialogue scenarios. Its core capabilities include:

Multi-step task execution: can initiate research, data organization, report generation, and PDF output within conversations

Multimodal input: supports voice, images, and documents, which can be uploaded for direct analysis and result generation

Personalized response styles: can be set to concise, structured, or conversational outputs

Offers two modes: Manus 1.6 Max (deep reasoning) and 1.6 Lite (high-speed tasks). Technically, the Max version is optimized for complex reasoning and creative tasks, while Lite is suitable for summaries, quick queries, and daily work scenarios.

Workflow Automation as a Main Application Scenario

Case studies released by Manus show that their target market includes knowledge workers and personal productivity scenarios, such as:

Setting up weekly automatic email and meeting summary preparations

Uploading photos to generate personalized avatars in different styles

Creating marketing video scripts and visual concepts based on product images

Additionally, users can directly create tasks or trigger automation flows via voice, making the agent a long-term personal assistant rather than a one-time Q&A tool.

Emphasizing Privacy and Task Isolation

Regarding user data security, Manus states that agents can only read content directly sent in private chats and cannot access other Telegram conversations, groups, or contact information. At the same time, a single request can coordinate multiple sub-tasks in the background, including research, code execution, and document generation, with delivery completed within the same chat thread.

This article about Meta’s Chinese AI startup launching Manus Agents highlights their low barrier to entry as a key advantage in competing against OpenClaw, first reported by ABMedia.