What Is Fellou?Fellou 是什么？
Fellou is a desktop AI browser built around agents rather than tabs. It launched on 11 May 2025 positioning itself as the world's first agentic browser, and on 2 September 2025 it released Fellou CE, a spatial agentic browser edition: work happens in a spatial interface where agents run in the background, scheduled tasks execute without supervision, and results arrive as interactive, visual elements rather than walls of text. The company claimed more than one million users at launch coverage — a vendor figure, not an audited count. The pitch is a shift from "chatting" to "executing": where a chatbot answers, Fellou is supposed to finish the job across websites and apps. Fellou's automation counterpart on this site is workflow automation — see Copy.ai for multi-step GTM workflows, though Copy.ai automates sales processes while Fellou automates browser work itself.Fellou 是以 Agent、而非标签页为核心的桌面 AI 浏览器。2025-05-11 首发时自称全球首个 Agent 浏览器；2025-09-02 又上线空间版 Fellou CE：在空间界面里让 Agent 在后台跑、定时任务无人值守执行、结果以可交互的可视化元素呈现，而不是大段文字。厂商在发布报道中称用户超 100 万——这是厂商口径，未经审计。它的立场是从“聊天”转向“执行”：聊天机器人只回答，而 Fellou 要跨网站、跨应用把活干完。本站的自动化对照见 Copy.ai（多步 GTM 工作流），不过 Copy.ai 自动化的是销售流程，Fellou 自动化的是浏览器操作本身。
The two modes that matter are Deep Search and Deep Action. Deep Search opens more than five shadow browsers in the background and searches public pages and login-required platforms (LinkedIn, X, Reddit, academic databases) in parallel, then synthesises a visual report with plans, comparisons and analysis — Fellou CE made Deep Search and visual reports free. Deep Action turns one instruction into a multi-step execution: fill a form, send an email, manage tabs, run a cross-platform workflow, with a preview of the credit (Spark) cost before it runs and human intervention possible mid-task. Around them sit Agentic Memory (a memory continuum across apps, so later tasks reuse earlier context), proactive intelligence (recommendations from behaviour), a smart timeline (return to any task context), local file integration by drag-and-drop, and the open-source Eko 2.0 framework for developers to build custom agents in natural language or JavaScript.真正干活的是两个模式：Deep Search 与 Deep Action。Deep Search 在后台开 5 个以上影子浏览器，跨公开页面与登录平台（LinkedIn、X、Reddit、学术库）并行搜，最后综合成带方案、对比与分析的可视化报告——Fellou CE 起 Deep Search 与可视化报告免费。Deep Action 把一句话变成多步执行：填表、发邮件、管标签页、跑跨平台工作流，执行前先预览积分（Spark）花费，任务中途人可介入。外围还有 Agent 记忆（跨应用的记忆连续体，后续任务复用先前上下文）、主动智能（按行为给推荐）、智能时间线（随时回到任一任务上下文）、本地文件拖放集成，以及供开发者用自然语言或 JavaScript 自建 Agent 的开源 Eko 2.0 框架。
Verdict: Fellou earns a trial when the bottleneck is browser labour itself — recurring cross-platform research, form-heavy workflows, or scheduled scans you currently do by hand. It does not earn a subscription on the strength of a chatbot comparison: judge it on one real multi-step job, timed against doing it manually, with the credit preview on. The credit model is the budgeting risk: workflow estimates (5 or 12 workflows per package) are rough guides, not guarantees, and complex jobs burn more. Start with the free Deep Search and visual reports, run one scheduled task, and only then price a package against your actual monthly workflow count.
实在的结论：Fellou 值得试的场景是瓶颈本身就是浏览器体力活——经常要做的跨平台调研、填表密集型流程、现在靠手工的定时扫描。别拿它跟聊天机器人比着买：拿一个真实多步任务、掐表跟手工对比、开着积分预览跑一遍，再决定。积分模式是预算风险：每个包“约 5 个或 12 个工作流”只是粗估、不是保证，复杂任务烧得更多。先用免费的 Deep Search 与可视化报告、跑一个定时任务，再按真实月工作流数去算包的价格。
— 中国吧 AI Tools Hub Editorial Team
Who Should Use It?谁适合用？
✓ Good Fit For适合以下场景
- Researchers and marketers doing recurring cross-platform research经常做跨平台调研的研究者与市场人员
- Anyone with form-heavy, repeatable browser workflows填表密集、可重复的浏览器流程使用者
- Developers who want custom agents via the Eko framework想用 Eko 框架自建 Agent 的开发者
✕ Not Ideal For不适合以下场景
- Users who only need answers, not executed tasks只要答案、不需要执行任务的人
- Windows-only teams until the Windows build is confirmed在 Windows 版确认前纯 Windows 团队
- Teams that need a published price list before procurement采购前必须看到公开价目表的团队
Getting Started快速开始
1. Download from fellou.ai and confirm the build for your OS1. 从 fellou.ai 下载，先确认对应你系统的版本2. Run one Deep Search free and check the visual report quality2. 免费跑一次 Deep Search，验可视化报告质量3. Time one Deep Action against manual work, then price a credit package3. 拿一个 Deep Action 跟手工掐表对比，再算积分包价格Key Features核心功能
- Deep Search: Parallel Research Into Visual ReportsDeep Search：并行调研成可视化报告 — Shadow browsers search public and login-required platforms in parallel and synthesise a visual report; free in Fellou CE.影子浏览器跨公开与登录平台并行搜索、综合成可视化报告；Fellou CE 起免费。
- Deep Action: One Instruction, Multi-Step ExecutionDeep Action：一句话、多步执行 — Fill forms, send email, manage tabs and run cross-platform workflows from natural language, with a Spark cost preview.用自然语言填表、发邮件、管标签页、跑跨平台工作流，执行前预览 Spark 花费。
- Scheduled Execution and Agentic Memory定时执行与 Agent 记忆 — Set recurring tasks to run unsupervised; memory carries context across apps and tasks.定时任务无人值守执行；记忆跨应用、跨任务携带上下文。
- Eko 2.0: Build Your Own AgentsEko 2.0：自建 Agent — Open-source framework to build intervenable agents in natural language or JavaScript, shareable as domain agents.开源框架，用自然语言或 JavaScript 搭建可介入的 Agent，还能分享为领域 Agent。
- Local-First, On-Device Execution本地优先、端侧执行 — CE positions execution on-device so sensitive data stays local; Fellou claims military-grade encryption and no behaviour tracking.CE 主张端侧执行、敏感数据不出本地；Fellou 宣称军用级加密、不追踪浏览行为。
Pros & Cons优缺点
✓ Pros优点
- Turns research and form work into delegated jobs, not open tabs把调研与填表变成可托付的任务、不是敞开的标签页
- Deep Search reaches login-required platforms, not just the public webDeep Search 能进登录平台，不只公开网页
- Scheduled execution and Eko custom agents cover recurring work定时执行与 Eko 自建 Agent 覆盖重复性工作
- Free Deep Search and visual reports in Fellou CEFellou CE 里 Deep Search 与可视化报告免费
✕ Cons缺点
- No verified official public list price — all figures are third-party核不到官网公开价——所有数字都是第三方口径
- Windows availability is unresolved across sourcesWindows 版是否可用的口径各家不一
- Credit consumption varies with workflow complexity; package estimates are rough积分消耗随工作流复杂度而变；包的“几个工作流”只是粗估
- Agent execution on logged-in accounts needs trust and supervision在登录账号上跑 Agent 执行，需要信任与监督
Use Cases应用场景
Popular ways people use it.常见用法：
📊 A Market Scan That Writes Itself自己写出来的市场扫描
Ask for startups and investors in a sector; get a visual report sourced across public and login platforms.问一个赛道的初创公司与投资人，拿到一份跨公开与登录平台取材的可视化报告。
📝 Forms and Outreach Without the Clicking不用点来点去的填表与触达
Delegate form filling, email sending and tab wrangling to Deep Action and review the result.把填表、发邮件、整理标签页交给 Deep Action，完事只审结果。
⏰ Recurring Scans on a Schedule按计划跑的定期扫描
Schedule market scans or report generation to run at set times without manual oversight.把市场扫描或报告生成排成定时任务，无人值守跑。
🧩 A Custom Agent for One Domain一个领域的专属 Agent
Build a domain agent with Eko 2.0 — law, design, healthcare — and share it with a team.用 Eko 2.0 给法律、设计、医疗等领域各做一个专属 Agent，再分享给团队。
Similar AI Tools相似 AI 工具
Related pages on this site: Copy.ai for multi-step workflow automation, and Cursor for agent-assisted development. If the job is only answering questions, a general chatbot is cheaper; Fellou's case is execution, not answers.本站相关页：Copy.ai（多步工作流自动化）与 Cursor（Agent 辅助开发）。如果工作只是问答，通用聊天机器人更便宜；Fellou 的立足点是执行、不是回答。