Manus Co-Founder Sets Record Straight on AI Agent Origins
In a pointed clarification, Yichao 'Peak' Ji, co-founder and Chief Scientist at Manus, has asserted that his company's newly launched personal agent mode has no technical connection to OpenClaw. This statement comes just days after OpenAI announced hiring Peter Steinberger, the solo creator of OpenClaw, sparking widespread internet comparisons between the two AI agent projects.
Timing and Context of the Clarification
The timing of Ji's post is significant. Manus rolled out its personal agent mode on Telegram shortly after OpenAI's acquisition of OpenClaw made headlines, leading to rapid online speculation about potential overlaps. Ji addressed this directly in a lengthy post on X, emphasizing the independence of Manus's technology.
"Manus personal agent mode is not based on OpenClaw. It is still built entirely on Manus's in-house architecture," Ji wrote. He acknowledged OpenClaw's influence but clarified it was philosophical rather than technical, crediting the project for helping shape industry thinking on trust and safety boundaries between users and autonomous agents.
OpenClaw's Philosophical Impact on Agent Safety
According to Ji, OpenClaw's real contribution lies in defining what he terms the "safety boundary"—addressing critical questions about how much autonomy an AI agent should have and when control should revert to the user. Manus had already grappled with these issues a year prior to its launch, implementing four key design choices from the outset to build user trust.
- Cloud Sandbox: Each task runs in an isolated virtual machine to prevent interference with the user's computer.
- Task Replay Mode: Allows users to safely explore agent actions in a read-only format.
- Interruptible Action Chain: Enables real-time inspection and correction of agent steps.
- GUI Takeover Option: Returns control to the user on demand during tasks requiring human intervention.
Ji noted that these features have since become standard across the agent industry, establishing a foundation of reliability.
Manus's Shift Toward Enhanced Accessibility
With trust now largely established, Ji revealed that Manus is pivoting toward greater accessibility in its new agent mode. The focus is on simplifying user interaction through natural language interfaces, smarter scheduling for parallel task execution, and a context-management system that maintains coherence across sessions. This move aims to lower barriers for non-technical users, aligning with Manus's mission to extend AI's reach beyond engineers.
Broader Industry Backdrop and Acquisition Dynamics
Ji's clarification unfolds against a charged backdrop of high-stakes acquisitions. OpenClaw transitioned from an open-source project to an OpenAI acquisition within weeks, with both Meta and OpenAI reportedly making billion-dollar offers for Steinberger. Meta, which acquired Manus for over $2 billion earlier this year, lost the bid for OpenClaw despite CEO Mark Zuckerberg's personal involvement in testing and providing feedback. Ji's statement appears aimed at distinguishing Manus's technology from OpenClaw's, underscoring the competitive landscape in the AI agent sector.
In summary, while OpenClaw has influenced philosophical discussions on agent safety, Manus maintains its technological independence, leveraging in-house innovations to drive the next wave of accessible AI tools.
