When AI starts to decide

Empowering Humanity: Manus AI’s Breakthrough in Agentic Intelligence Unveiled at Web Summit Lisbon 2025

At Web Summit Lisbon 2025, Mr. Steve Clemons interviewed Mr. Tao Zhang, Co-founder and CPO of Manus AI, on agentic AI. Manus AI leads in autonomous systems, where AI decides and acts. Mr. Clemons noted Manus AI’s practical implementation, distinguishing it from other major firms. Mr. Zhang explained “Manus” is Latin for “hand,” reflecting their focus on AI’s tangible impact in the physical world, complementing the “brain” aspect of AI development.

Manus AI’s technical innovation is its general agent system, operating without predefined workflows. The AI model intelligently predicts and executes actions, utilizing fundamental tools like file editing, command execution, and web browsing. This enables autonomous planning and completion of diverse tasks. Since its March 2025 launch, Manus AI has achieved millions of active users and over $100 million in annualized revenue, demonstrating rapid market adoption.

Mr. Zhang detailed the platform’s continuous evolution, including the recent Manus 1.5 release. This update significantly enhanced web app building, allowing users to create AI apps with natural language in under 20 minutes, and expanded capabilities for handling images and videos. He shared personal insights: his wife uses Manus AI extensively for stock trading research, informing her investment decisions.

He personally leverages Manus AI each morning to prepare for meetings, checking his Google Calendar and gathering background information on external attendees from LinkedIn or personal blogs. This ensures thorough preparation. Inspiring real-world applications include an NGO in Brazil using Manus AI to build a vital website, providing resources for parents of children with autism.

Another example featured a 7-year-old Japanese boy who used Manus AI as an interactive environment to learn Python, showcasing its potential as an accessible educational tool. Addressing concerns about autonomous AI, Mr. Zhang emphasized Manus AI’s commitment to transparency. The product clearly displays the agent’s actions and thought processes, fostering user trust.

Users can intervene at any time to provide new instructions or alter the agent’s approach. For sensitive actions, such as sending emails or making purchases, the agent automatically pauses, requiring explicit user permission before proceeding. This ensures user control and safety, aligning with a philosophy of enhancing human capabilities rather than replacing them.

Mr. Zhang shared how Manus AI helps him manage ADHD by taking his numerous ideas and initiating progress, fostering a greater sense of accomplishment and confidence. Looking ahead, he identified proactivity as the next major frontier. Future agents will anticipate needs and complete tasks even before being explicitly asked, moving beyond reactive commands.

He described an internal prototype where Manus AI checks his schedule and email for undone tasks each morning, completing them before he wakes up. The ultimate metric for Manus AI, Mr. Zhang concluded, is “agentic hours per user.” This emphasizes the time the agent spends working autonomously in the background, highlighting continuous, seamless AI assistance that deeply integrates into daily life.