The Singaporean AI startup Manus has launched an experimental feature called “Wide Research” that allows users to run more than 100 AI agents in parallel to complete large-scale tasks. According to an article by Carl Franzen in VentureBeat, this approach contrasts with the “Deep Research” methods used by competitors like OpenAI and Google, which typically rely on a single agent for extensive, in-depth analysis.
Manus states its system can tackle complex data analysis and creative projects simultaneously. In a demonstration, the company used its agents to compare 100 different sneakers based on design, price, and availability, delivering a sortable spreadsheet within minutes. The company says the technology can also generate creative assets, such as poster designs in 50 distinct visual styles at once.
The feature is based on parallel processing, where each subagent is a fully capable instance of the Manus platform rather than a specialized one. This architecture is designed to allow for flexible and scalable workflows.
However, the VentureBeat report notes that Manus has not provided performance benchmarks or technical comparisons to prove that this parallel approach is more effective than single-agent systems. The practical benefits of “Wide Research” remain unproven based on the currently available information.
The feature is available now for subscribers of the company’s “Pro” plan and is expected to be rolled out to other tiers over time.