Fellou

Fellou is best understood as a agentic AI browser automation product for researchers and operators who delegate browser work. I fetched the vendor homepage and pricing page with curl before writing this profile, then used the visible page text plus the staged product data. The goal is not to repeat a marketing tagline; it is to explain what a builder can actually test. The current evidence points to a tool that should be evaluated around a narrow operational workflow, not a vague promise that AI will save time everywhere.

Pricing evidence: Free ($0/month): 1000 Sparks new-user bonus, about 4 tasks, 2 scheduled tasks, unlimited concurrent tasks during limited-time offer; Plus ($19/month): 2000 Sparks/month, about 8 tasks, 3 scheduled tasks, priority support, early access; Pro ($39.90/month): 5000 Sparks/month, about 20 tasks, 5 scheduled tasks, priority support; Ultra ($199.90/month): Unlimited Sparks during limited-time offer, unlimited scheduled tasks, exclusive priority support; annual shown at $1,920. This matters because AI tools in this category often hide the real constraint in credits, exports, retention, app builds, seats, or support. Before procurement, confirm the billing unit, overage behavior, cancellation terms, data retention, export rights, and whether collaboration features are included or paid add-ons.

The concrete feature set includes Deep Action agent, deep search across multiple sources, reviewable action plan before execution, scheduled tasks, browser use and computer use, visual report, image, music, and code agents. These are specific enough to design a real pilot. Avoid testing with a toy prompt. Use representative inputs, edge cases, and the kind of messy source material or production traffic the team already handles. If the tool cannot handle those cases with human review, it is not ready for a critical workflow even if the demo looks impressive.

Best use cases include competitor pricing research, market update reports, cross-site data collection, personal browser automation. A practical pilot should define the input, expected output, reviewer, pass/fail bar, and time budget before the first run. For example, run 10 to 20 real tasks, record the number of correction loops, measure time saved, and decide whether the output is good enough to ship, good enough to edit, or only useful as inspiration. Also test failure behavior: bad prompts, missing data, malformed files, slow websites, or ambiguous requirements.

Pros: the tool has a clear category fit, visible features that can be tested, and enough information to compare against adjacent products. It should be useful when speed to first draft, workflow visibility, or prototype creation is more important than perfect automation. Cons: humans still need to review outputs, teams must verify privacy and retention, and the value depends heavily on input quality and process discipline. If pricing is unverified, do not assume a free tier, export access, or production rights.

What makes Fellou different is its category focus: it is not just a chatbot bolted onto a generic interface. The strongest buying case is when that focus maps directly to a painful existing workflow. Compare it with related internal options before committing: Browser Use (/tools/browser-use), Browserbase (/tools/browserbase), Apify (/tools/apify), AgentOps (/tools/agentops).