Manus Launches ‘Wide Research’ AI with 100+ Web-Scouring Agents
▼ Summary
– Manus, a Singaporean AI startup, launched “Wide Research,” a feature enabling large-scale tasks using parallelized AI agents (100+ at once) for research or creative work.
– Unlike rivals focusing on deep, single-agent research, Manus emphasizes wide-scale parallel processing for tasks like comparing 100 sneakers or generating diverse design styles.
– Wide Research is available on Manus Pro ($199/month) and will roll out to lower-tier plans, offering varying credit limits and concurrent task capabilities.
– The company claims Wide Research leverages a flexible, general-purpose agent architecture but lacks performance benchmarks to prove its advantages over sequential approaches.
– Manus acknowledges Wide Research is experimental, facing industry-wide challenges like subagent coordination and reliability seen in similar AI systems.
Singapore-based AI startup Manus has unveiled an innovative approach to large-scale research and creative tasks with its new “Wide Research” feature. Unlike competitors focusing on deep, single-agent analysis, Manus leverages parallel processing by deploying over 100 AI agents simultaneously to tackle complex projects. This method promises faster, more diverse outputs, from product comparisons to design exploration, by distributing workloads across specialized subagents.
The system operates on Manus’s proprietary multi-agent platform, where each subagent functions as a fully independent instance rather than a predefined role. In a demonstration, the company showcased how 100 agents could analyze sneaker designs, prices, and availability in minutes, compiling results into sortable spreadsheets or webpages. Another example highlighted 50 agents generating distinct poster styles, delivering polished designs in a downloadable package.
Pricing tiers make Wide Research accessible to different user needs, starting with a free plan offering limited capabilities. Paid subscriptions range from $19/month for basic features to $199/month for advanced tools, including early access to beta functionalities. Annual payments come with a 17% discount. Currently, the feature is available to Pro subscribers, with gradual rollouts planned for lower-tier users.
While the technology demonstrates architectural ambition, questions remain about its efficiency compared to traditional sequential processing. Manus hasn’t provided benchmarks quantifying speed or accuracy gains, leaving the practical advantages open to scrutiny. Industry observers note that subagent systems elsewhere, like Claude’s Code, have faced criticism for high resource consumption and coordination issues.
The launch reflects Manus’s broader vision of AI as a scalable, cloud-based workflow solution. By treating each session as a dedicated virtual machine, the platform aims to streamline complex tasks through natural language commands. Whether this approach outperforms established methods will depend on real-world adoption and refinement. For now, Wide Research offers a glimpse into how parallelized AI could reshape research, creativity, and data analysis at scale.
Editor’s Note: A previous version of this article misstated Manus’s headquarters location and model dependencies. These inaccuracies have been corrected.
(Source: VentureBeat)
