Manus

Manus is the AI agent platform that exploded onto the scene in early 2025 from a Chinese AI startup called Butterfly Effect, and quickly became one of the most-discussed AI products of the year. The pitch: a general-purpose AI agent that doesn't just answer questions like ChatGPT — it actually goes off and does multi-step tasks autonomously, browsing the web, running code, manipulating files, and producing real outputs while you watch (or do something else). The demo videos when Manus launched were the kind of thing that made people say 'this changes everything' for a week. Tell it 'analyse the top 10 SaaS companies in this category and write me a competitive landscape brief', and Manus opens a browser, hits Google, navigates to each company site, reads pages, takes screenshots, runs Python to analyse data, writes a document, and delivers it in 20-30 minutes — all in a visible workspace you can monitor in real time. It's the most concrete version of 'AI agents' that consumers have been able to try. For founders the use cases that actually work in practice: • Competitive research (the killer use case — produces decent first-pass landscapes) • Lead enrichment (give it a list of company URLs, get back structured data) • Data scraping for one-off tasks (rather than building it yourself with Apify) • First-draft research reports (financial, market, technical) • Code prototypes for self-contained scripts The reality, after the initial hype: Manus is genuinely impressive but inconsistent. About 50-70% of tasks complete usefully; the rest fail mid-execution, take much longer than expected, or produce something that needs significant cleanup. For tasks where you can't easily verify the output, you have to spot-check thoroughly. It's a productivity tool, not a magic 'I never have to do this again' button. The pricing is access-based. Initially the product launched invite-only with waitlists. As of late 2025 it's available with tiered subscriptions starting around $39/month for Starter access (limited daily credit budget) and Pro at $199/month for heavier users. Enterprise tiers exist for businesses. The exact pricing has shifted multiple times — check the current page when you sign up. The geopolitical context matters. Butterfly Effect (the company behind Manus) is based in China. Some Western enterprises have hesitated to use it for sensitive workloads given data residency and export-control concerns. For indie founders doing general research the concern is mostly theoretical, but if your use case involves company-sensitive data, evaluate accordingly. My take: try Manus if you do a lot of multi-step research or repetitive web-based workflows. The 50-70% success rate is good enough for the cost when the alternative is doing the work yourself. For mission-critical tasks where you need 99%+ reliability, supervised use of Claude / ChatGPT with explicit prompting wins. Manus is best for 'tasks I'd skip if I had to do them manually' — that's where the agent earns its place.