FellouTranslation site

1. Core positioning: From “browsing tools” to “Agentic execution entrance”, define a new form of browser

The key difference between Fellou and traditional browsers lies in its positioning of “AI Agent and browser deep integration” – not only can display web pages, but also understand intent, disassemble tasks, and execute autonomously, solving productivity pain points through three core features:

(1) Hybrid shadow space: human-machine partition cooperation, which is more controllable without interfering with each other

Fellou pioneered the physical isolation mode of “foreground browsing + background execution”, completely solving the chaotic problem of “AI and manual screen grabbing”:

• Dual space parallel operation: Users can scroll through web pages and watch videos normally on the main interface of the front desk, while the AI Agent performs tasks in the independent “shadow space” background, without the need to open additional tabs or software. For example, the user command “Organize NVIDIA’s latest financial report and generate PPT” can continue to browse other content, and Fellou will automatically complete web page crawling, data extraction, and PPT production in the shadow space, without interfering with the front desk operation throughout the process.
• Real-time and controllable execution process: The shadow space status is fully transparent, and users can check the progress of tasks at any time (e.g., “financial report data extraction→ chart generation→PPT layout”), and can immediately pause, modify instructions, or terminate tasks if deviations are found, avoiding the risk of AI “deviation”. A user reported that Fellou found source deviations when crawling competing product data, and promptly stopped adjusting keywords to ensure accurate results.
• Cross-System Compatibility: Supports running on all platforms such as Windows, macOS, and Linux, providing a consistent “foreground + backend” collaboration experience on both desktops and laptops, with compatibility far exceeding similar tools.

(2) Deep action engine: From “instruction response” to “intention execution”, complex tasks can be done with one click

Fellou breaks through the limitations of traditional AI “step-by-step disassembly of instructions” and realizes the closed loop of “intent input → result delivery” through intelligent task scheduling:

• Automatic task disassembly and collaboration: In the face of vague or complex requirements, Fellou will first intelligently ask for clear core intentions (for example, if the user says “I will travel to Japan next week”, it will actively ask about destinations, budgets, transportation preferences, etc.), and then disassemble the requirements into task trees, and send special agents such as login, search, and organization to execute them in parallel. For example, the instruction “Find 10 jobs that match my LinkedIn profile and apply” will simultaneously launch “Resume Analysis Agent→ Job Search Agent→ Fill in the form →Delivery Tracking Agent”, without the user manually operating each step;
• Outstanding cross-site automation capabilities: The success rate of “write tasks” (login, form filling, etc.) in the Web Bench benchmark reached 72%, far surpassing Manus (60%) and OpenAI (59%). It supports automatic login to LinkedIn to search for talents, fill in web forms, and synchronize content across platforms, and an HR uses it to batch screen engineer resumes, increasing efficiency by 4 times.
• Cost Transparency and Privacy Protection: The required “sparks” (corresponding tokens) costs will be clearly displayed before executing tasks to avoid hidden consumption. Sensitive operations (such as login information and resume content) are processed locally and not uploaded to the cloud, taking into account efficiency and security.

(3) Multi-modal understanding and traceability: graphic and text collaboration is more professional, and the information is credible and verifiable

Fellou version 2.2.0 achieves a leap in multimodal capabilities, and at the same time strengthens information traceability, making the output results more professional and credible:

• Graphic and text collaborative analysis: For the first time, it supports the understanding and crawling of web images, whether it is product graphics, data charts or movie posters, it can accurately identify the content. For example, the command “Search for posters at the annual Cannes Film Festival and make a report” will automatically crawl the poster images, extract design elements, theme styles and other information, and generate an analysis report with pictures and texts, without manual screenshot stitching;
• Full data source traceability: The report generated after batch search will be accompanied by links to all original web page sources, and users can directly jump to verify the authenticity of the information. A market researcher used it to sort out industry data and quickly verify 3 data deviations through the traceability function to ensure the credibility of the report.
• Multi-format export adaptation: Supports exporting results to various formats such as TXT, MD, HTML, CSV, XLSX, PPTX, etc., adapting to different scenarios such as report writing, data processing, and presentation reports without the need for additional format conversion.

2. Functional matrix: Focusing on “Agentic execution”, create a full-scenario intelligent toolset

Fellou’s functional design closely follows the three goals of “efficient, controllable, and professional”, and each module has been cross-verified by official information, which is 100% consistent with the measured experience:

(1) Core execution function: cover the needs of multi-scenario tasks

• Deep Search and Content Generation:
  • Intelligent research: Conduct multi-source information crawling for any topic (such as “AI Agent development trend”), integrate industry reports, news, and academic literature to generate structured reports with core conclusions and data support.
  • Content creation: Assists in generating textual content such as articles, scripts, and stories, supports one-click publishing to social media, and marketers can use it to create promotional copy in batches, increasing efficiency by 5 times.
  • Code and Tool Support: Developers can build custom agents through the Eko framework, supporting the generation of utility code such as World of Warcraft macros to solve duplicate development problems.
• Cross-platform automation:
  • Workplace Office: Automatically log in to your email address to send emails, manage schedules, fill in reimbursement forms, and support LinkedIn job search and resume delivery.
  • Data processing: Crawling web data to generate csv/xlsx files, automatically calculating financial data (income, expenses, profits) and generating analysis charts to assist in financial decision-making;
  • Creative design: Support image editing and painting creation assistance, and designers can get inspiration through the command “Generate Techno-style product sketch” and then manually optimize it.
• Multimodal Reporting and Traceability:
  • Graphic report generation: Automatically extract web page images and insert them into reports, support custom typesetting style (professional/creative), and can be directly exported to PPTX for reporting after generation.
  • Traceability verification: The report comes with the original web link, and clicking can jump to the verification information, which is suitable for scenarios that require high credibility, such as academic research and market analysis.

(2) Space and task management function: flexibly control the execution process

• Hybrid Shadow Space Management:
  • Space switching: Use shortcut keys to quickly switch between front-end browsing and shadow space to view task progress or adjust commands.
  • Multi-task parallelism: Support multiple shadow spaces to perform different tasks at the same time (such as “sorting out financial reports” and “searching for jobs” in parallel), and the tasks do not interfere with each other.
• Task life cycle control:
  • Progress tracking: Each task displays “Percent Completed + Current Steps”, such as “Resume Delivery: 80%, 7th Job Feedback is Being Tracked”;
  • Operational intervention: Support pausing, restarting, and editing tasks, such as finding search keyword deviations, you can pause and modify the keywords before continuing execution;
  • History Backtracking: All task results are automatically archived, and history can be searched by time and type, which is convenient for subsequent reuse or modification.

(3) Version and pricing: hierarchical adaptation, taking into account experience and professional needs

Fellou adopts a tiered model of “free trial + subscription + customization”, with clear division of benefits and is completely consistent with the official information:

3. Usage process: Four steps to start intelligent execution, even if you have zero foundation

Fellou’s workflow follows the logic of “simple input→ background execution→ controllable adjustments→ result delivery”, which is exactly in line with the official “Getting Started” guide:

(1) Step 1: Download and install, log in to your account

1. Get the installation package: Visit the Fellou official website (https://fellou.ai), click the “Download” button, select the Windows or macOS version according to the system (Linux version is supported by synchronization), and download without an invitation code;
2. Quick login: After the installation is completed, register through your email or log in directly to enter the main interface (the left side is the command bar, the right side is the browsing area, and the bottom is the task timeline).

(2) Step 2: Enter the intention and clarify the needs

1. Natural language instructions: Describe requirements in everyday language in the command bar, without technical terms, and can include task objectives, output requirements, etc. For example:

• • Research scenario: “Crawl 10 core documents in the AI Agent industry in 2025, generate a review report with references, and export it to docx format”;
  • Workplace scenario: “Help me compare the prices of MacBook Air from Amazon and JD.com, generate a comparison table and send it to my work email”;
  • Life scenario: “Develop a 2-week fat loss fitness plan with daily training content and dietary recommendations, generate PDF guides”;

2. Requirements confirmation: If the instructions are vague, the Fellou will proactively ask for details (such as “Does the fitness plan be suitable for beginners or basic people?”). ), and then generate the task execution plan and display the cost.

(3) Step 3: Start the task, monitor and adjust

1. Background execution: After clicking “Start”, the task will automatically enter the shadow space to execute, and the main interface can continue to browse the web without being disturbed.
2. Progress management: View the task progress through the bottom timeline, and click on the task to expand the details (e.g., “Crawling Documents→Part 3 / 10 of 10”);
3. Real-time intervention: If you find a problem, you can click “Pause”, enter a modification instruction (e.g., “Literature is limited to publication in the past 6 months”), and restart the task after confirmation. High-risk operations (such as logging in to an account) will pop up to request authorization to ensure security.

(4) Step 4: Get the results, export and reuse

1. Result preview: After the task is completed, the shadow space automatically pushes the results, and supports online preview of reports, tables, PPTs, and other content.
2. Export and Share: Choose the desired format (such as pptx, csv) to export to the local area or send it directly via email; Professional users can sync to the team collaboration platform;
3. Traceability and Verification: Click on the source link in the report to jump to the original webpage to verify the information and ensure the content is trustworthy.

4. Application scenarios: covering diverse groups of people and implementing the value of all types of productivity

The functional design of Fellou accurately matches the needs of different user groups, which is highly consistent with official cases and measured experiences:

(1) Academic and research scenarios: Efficiently complete research and reports

Literature review and analysis:

• Requirements: “Search for core papers in the field of AI large model security in the past year, extract research methods, conclusions, and generate a review with charts”;
• Fellou Action: Crawl IEEE Xplore, arXiv and other platform literature in shadow space, identify core data to generate trend charts, automatically mark reference sources, and complete the original workload of 1 day in 2 hours.
• Results: The researcher’s literature collation efficiency has been improved by 80%, and the credibility of the report has been recognized by peers through the traceability function.

(2) Workplace and office scenarios: Simplify cross-platform operations

Recruitment and Talent Search:

• Requirements: “Search for engineers with more than 5 years of AI Agent development experience on LinkedIn, compile 10 high-quality resumes and mark core skills”;
• Fellou Action: Automatically log in to LinkedIn to perform keyword searches, filter qualified candidates, extract skills and project experience to generate a structured form, and synchronously mark resume source links;
• Results: HR reduced resume screening time from 8 hours to 1.5 hours, improving accuracy by 30%.

(3) Marketing and creation scenarios: automated content and data tracking

Social Media Operations:

• Requirements: “Investigate the social media promotion strategy of competing products Q3, generate analysis reports and create 3 promotion copies suitable for Xiaohongshu”;
• Fellou action: Crawl the content of competing Weibo and Xiaohongshu, analyze the publishing frequency, style, and interactive data, generate graphic reports, and then create customized copywriting according to the report conclusions, supporting one-click publishing;
• Results: Marketers are 5x more efficient in content creation and research, and 40% more adaptable copywriting compared to manual creation.

(4) Personal life scenarios: customized planning and services

Fitness and travel planning:

• Requirements: “Create a 4-week fat loss plan for beginners with a weekly training plan, diet checklist, generate a printable PDF guide”;
• Fellou action: After asking about weight, exercise basics and other information, integrate professional fitness website resources to generate daily training steps (including links to action diagrams) and dietary suggestions, and automatically export them as PDFs;
• Results: Personal planning time has been reduced from 3 hours to 20 minutes, and the professionalism of the program has been verified by traceability fitness website information.