Fellou Review: Agentic Browser That Does Instead of Just Shows
Fellou browser review: an agentic AI browser that finds sites, fills forms, and automates tasks. Features, pricing, limits, and how it stacks up against Claude, Google’s browser agents, Comet, Dia
What if your browser could act for you instead of just showing pages? That’s what Fellou aims to do — an agentic AI browser built for autonomous browsing and browser automation. In this Fellou review, I tested a real workflow with two tasks. Before I share results, let’s understand what Fellou is.
No Time to Read? Here’s the Scoop
Agentic Browser – Fellou doesn’t just load pages, it acts on your instructions across sites.
Form Automation – I used it to find newsletter submission sites, scrape my site (thetoolnerd.com), and auto-fill forms. Worked well on simple pages.
Smart Memory – Stores scraped info locally, so it can reuse data without repeating steps.
Not Perfect – Gets stuck on complex layouts, loops at times, and can’t bypass CAPTCHAs.
Pricing – Credit model (Sparks). Free plan to try, paid tiers at $19, $39.9, and $199 with higher limits.
Competition – Faces stiff rivals like Claude’s Computer Use and Google’s upcoming Chrome agent, Perplexity Comet etc..
Verdict – Promising tool with clever features, but not yet reliable enough for fully hands-off automation.
What Fellou Is: Agentic AI Browser
Fellou calls itself an agentic browser. It functions as an AI web agent and an autonomous browser for routine workflows. Instead of passively loading pages, it can:
Execute multi-step tasks from plain instructions
Work across sites and apps
Store and reuse scraped data
Run actions in a “shadow” workspace
Give credit (Spark) estimates before execution — very good feature
If you have read my previous article about Perplexity Comet and Dia Browser, Fellou is somewhere in line with Perplexity Comet.
In the Online-mind2web benchmark, the task completion rate has increased from 31% to 80%.
My Task & Experience
To evaluate Fellou’s capabilities, I tested it with two distinct tasks:
TASK 1 : YC AI Startup Research
I asked Fellou to extract a list of Y Combinator AI startups from the past year, including company names, websites, and descriptions.
The results were impressive. Fellou successfully:
Navigated multiple websites including Y Combinator’s own site
Scraped information from Reddit and other platforms
Compiled a comprehensive list of AI companies backed by YC
Organized the data in a structured format with all the requested details
This task showcased Fellou’s ability to perform complex research across multiple domains and synthesize information without constant supervision.
TASK 2 : Find sites where I can submit my newsletter, scrape my own website for details, and auto-fill the forms (form automation)
Here’s how it went:
Discovery: It actually browsed Reddit, blog directories, and Google results, picking out pages that looked like newsletter submission forms. This part felt almost like having a junior researcher scanning the web for me.
Scraping: Before filling forms, it visited my site (thetoolnerd.com) and pulled the required details like name, URL, and description. I didn’t have to feed it manually — it figured out what was needed. It also scraped my LinkedIn profile to get more details about me. I gave the url to it.
Form Filling: On simple, well-structured forms, it filled and submitted without much trouble — a solid example of browser automation and form autofill.
Memory: The best part — it cached the data locally. So when I ran the task again, it reused the info without scraping my site all over again. That saved both time and credits.
Stumbles: On more complex pages, it struggled. At times it got stuck repeating the same step, or failed when the page layout was dynamic. CAPTCHA pages were a complete blocker.
Overall, the task showed me both its strength (automation with memory) and weakness (unreliable on tricky sites). For repetitive work like outreach, it’s promising but not hands-off yet.
Other Use Cases
Beyond form-filling, Fellou can handle various complex tasks:
Think of it as light, no-code web automation for repeatable tasks, and web scraping automation when pages are simple and structured.
Research & Analysis: Conduct market, academic, and competitor research across platforms (including LinkedIn, X, Quora) with shareable visual reports. Monitor products, gather pricing data, and compile structured reports.
Content Generation & Publishing: Create multi-channel content from blogs, schedule social updates, post directly to platforms like LinkedIn or X, and transform existing content into different formats.
Business Operations: Automate sales prospecting (find leads, enrich profiles, draft outreach), recruiting (source candidates, summarize profiles), and email management (triage inbox, draft replies).
E-commerce & Shopping: Search products across sites, compare options, add items to cart, and capture screenshots for later reference.
Data Management & Workflow: Scrape and reuse data across submissions, tag and organize information, automate reporting, and run background tasks while you work on other things.
Limitations
Loops and task failures on complex pages
Can’t bypass CAPTCHAs or human checks sometimes
Costs can vary based on workflow length
Still early and less reliable.
Pricing (Fellou pricing, Sparks)
Fellou pricing and cost: it runs on a credit model (Sparks):
Free plan with limited Sparks
Paid tiers ($19, $39.9, $199) offering more Sparks, scheduled tasks, and support
Each workflow shows an estimated Spark cost before you run it
The Bigger Picture
I’ve earlier compared Perplexity Comet vs Dia browser — Fellou feels like the next step. But it faces stiff competition: Claude’s new Computer Use and Chrome extension, and Google’s rumored Chrome agent (Project Jarvis). These players could quickly dominate this space.
Conclusion
Fellou is promising — a browser that acts, not just displays. It has smart features like memory and credit transparency but needs more reliability. Worth exploring, especially if you want to cut down on repetitive web tasks.