Installation
Gatsbi: Innovative Research AI Co-Scientist and Paper Writer
Gatsbi desktop application is designed for Windows and macOS that focuses on creativity, privacy, and productivity. All your creative work stays local on your device, ensuring maximum security and privacy.
Desktop Application Advantages
- All your creative work stays local on your device
- Servers do not collect or store user input/results
- Uses third-party AI service providers securely
- Maximum privacy and data protection
System Requirements
| Component | Minimum Requirement |
|---|---|
| Operating System | Windows 10/11 (64-bit) or macOS Sequoia 15.1+ (Apple Silicon) |
| RAM | 4 GB |
| Processor | 2.5 GHz × 2 cores |
| Storage | 2 GB disk space |
| Display | 1440 × 900 resolution |
Web Version
A beta web version is available for Pro users, accessible from the user space left menu. This provides flexibility to use Gatsbi from any device with an internet connection.
Ready to get started? Download Gatsbi from the official website and begin your research innovation journey.
First Boot Configuration
Setting up Gatsbi for the first time involves logging in and configuring your AI service provider.
Login Methods
1. Direct Login
- Enter your email address
- Enter your password
- Optional: Check "Remember my password" for convenience
2. Third-Party Account Login
You can also login using:
- Google Account
- LinkedIn Account
- GitHub Account
AI Service Provider Configuration
Hybrid (Default)
Orchestrates multiple open-source AI models for optimal performance. No additional configuration required. An alternative option is available for regions where the default service cannot be accessed.
OpenAI
Use OpenAI's powerful models. Requires personal OpenAI API key. Option to configure proxy (base URL) for restricted regions.
Setting Up OpenAI
- Obtain your OpenAI API key from platform.openai.com
- Enter the API key in Gatsbi settings
- Configure proxy (base URL) if needed (for regions with OpenAI access restrictions)
- Test the connection to ensure proper setup
Tip: Start with the Hybrid provider for immediate use. You can switch to OpenAI later if needed.
Gatsbi Innovator
AI-Powered Research Innovation and Idea Generation
Gatsbi Innovator is the core ideation engine that helps researchers generate novel ideas, develop research proposals, and create comprehensive academic outputs. It combines advanced AI with systematic research methodologies to produce innovative solutions.
Workflow Overview
- Topic Input: Enter your research area or problem statement
- Systemic Analysis: AI performs comprehensive background research
- Idea Generation: Receive scored, referenced innovation candidates
- Idea Expansion: Develop selected ideas into full implementations
- Output Generation: Create paper manuscripts or patent disclosures
Available Functions
- Start Ideation & Best Practices - Begin your innovation journey and optimize results
- Research Paper Drafting - Generate academic manuscripts
- Patent Disclosure Generation - Create IP documentation
Getting Started: Select "Gatsbi Innovator" from the service menu to access all ideation and generation features.
Start Ideation & Best Practices
Unlock Innovative Ideas with Gatsbi Innovator and Optimize Your Workflow
This comprehensive guide covers both the ideation process and best practices for using Gatsbi Innovator effectively.
Part 1: Starting Your Innovation Journey
Step 1: Select Gatsbi Innovator
Ensure "Gatsbi Innovator" is selected in the top-left service menu to activate the ideation engine.
Step 2: Enter Your Research Topic
Best Practice: Use specific topics for better results (e.g., "interpretable large language models" instead of just "large language models"). The system provides smart recommendations if your input is too broad.
Step 3: Systemic Analysis
Gatsbi performs multi-step automatic analysis before idea generation:
- Component Analysis and Research Gap Identification
- Technology Trend Assessment
- Feasibility Evaluation
- Innovation Potential Scoring
Step 4: Idea Generation
The system generates 10-20 idea candidates, each including:
Scoring Metrics
1-5 star rating
Comprehensive evaluation of innovation and feasibility
References
Up to 10 relevant citations
Supporting literature
Description
Brief overview
Potential impact analysis
Step 5: Idea Expansion
Expand your selected generated idea into comprehensive implementation details. The expansion provides:
- Modeling Approaches: Mathematical formulations, algorithms, and theoretical frameworks
- System Architecture: Module integration relationships and component interactions
- Implementation Methodology: Step-by-step development process and technical requirements
- Technical Specifications: Detailed parameters, constraints, and performance metrics
- Integration Strategy: How different system modules connect and communicate
The expansion transforms a high-level concept into an actionable research plan with concrete technical details.
Part 2: Best Practices & Optimization Tips
Input Optimization Strategies
For Optimal Results:
- Be Specific: Use precise technical terminology
- Include Research Target and Improvement Direction: Clearly specify what you want to improve and how
- Include Context (if necessary): Mention your field and application area when relevant
- Intervene in Analysis Process (if needed): Adjust intermediate analysis steps as described in the next section
Working with Automatically Generated Analyses
Control Buttons for Each Analysis Step
Every generated section includes three powerful control buttons:
Copy Button
Exports text in markdown format for easy integration into other documents
Regenerate Button
Creates alternative versions of current and subsequent sections
Regenerate with Instructions
Intervene in the analysis process with custom prompts
Advanced Techniques
1. Iterative Refinement
- Focus on core components to narrow down your research subject
- Use "Regenerate with Instructions" to focus on promising directions
- Review and evaluate generated ideas for quality and relevance
2. Multilingual Capabilities
- Input topics in any language
- Core analysis in English ensures accuracy
- Comments translated back to input language
3. Quality Assurance Checklist
- ✓ Verify technical accuracy of proposals
- ✓ Check citation relevance and recency
- ✓ Ensure feasibility aligns with resources
- ✓ Validate novelty claims against literature
Common Pitfalls to Avoid
- Too Broad: Avoid generic topics like "machine learning"
- Too Application-Oriented: Focus on methodological innovation rather than pure applications
- No Specific Research Target: Define clear research objectives or directions
- Wrong Tool Selection: Use Gatsbi Writer and Reviewer for non-methodological research
Pro Tips for Maximum Productivity
- Focus Your Research: Try focusing on specific systems or modules to find the most suitable innovation points
- Monitor Analysis Process: Read intermediate analysis steps; use "Regenerate with Instructions" to intervene if results deviate from expectations
- Systematic Exploration: Expand top 3-5 ideas before deciding; compare their technical feasibility and innovation potential
- Leverage References: Check cited papers for additional insights, related work, and potential research gaps to explore
Remember: The quality of output depends on the quality of input. Take time to craft your research topic thoughtfully, and don't hesitate to iterate!
Research Paper Drafting
AI Paper Writing with Citations, Figures, and More
After generating ideas, Gatsbi Innovator seamlessly connects to Gatsbi Writer to complete paper manuscripts or patent disclosures with one click.
Key Features
- One-click research paper generation
- Automated creation of all manuscript elements
- Automatically retrieves and inserts real citations
- Multiple export formats
Automated Elements
Citations & References
Real automatic in-text citations and comprehensive reference list generation
Visual Elements
Professional figures, charts, and diagrams
Mathematical Content
Properly formatted equations and formulas
Data Presentation
Experimental tables and data visualization
Manuscript Generation Process
Step 1: Configure Settings
- Select word count (up to 5,000 words)
- Use word count estimator tool (if unsure)
Step 2: Generate Manuscript
Click "Write a Paper Manuscript" button to start automated generation.
Step 3: Review Generated Content
The system creates comprehensive manuscript elements:
- Title and Abstract
- Introduction with literature review
- Methodology section
- Results with figures and tables
- Discussion and implications
- Conclusion
- References
Export Options
| Format | Best For | Features |
|---|---|---|
| DOCX | Journal submissions | Full formatting, easy editing |
| LaTeX | Academic publications | Professional typesetting |
| Markdown | Version control | Plain text, portable |
Supported Citation Styles
| Citation Style | Common Use | Field |
|---|---|---|
| APA | Psychology, Social Sciences | General academic writing |
| IEEE | Engineering, Computer Science | Technical papers |
| Harvard | Business, Economics | International publications |
| Chicago NB | History, Literature | Humanities research |
| AMA | Medicine, Health Sciences | Medical journals |
Unique Advantages
- Advanced language modeling for natural writing
- Minimal user intervention required
- Maintains academic writing standards
- Consistent formatting throughout
Transform your ideas into professional academic papers with expert guidance!
Patent Disclosure Generation
Create comprehensive patent disclosure documents to protect your intellectual property.
As an alternative to a paper manuscript, you can transform your innovation into patent documentation.
What is a Patent Disclosure?
A patent disclosure document is a detailed technical document describing an invention. Its purpose is to:
- Protect intellectual property
- Establish novelty of an innovation
- Provide technical documentation for patent filing
- Create a formal record of invention
Generation Process
Step 1: Switch to Patent Mode
Change from "Research" to "Patent" mode in the application.
Step 2: Select Language
Choose from 10 supported languages:
- English
- French
- German
- Chinese Simplified
- Chinese Traditional
- Japanese
- Korean
- Spanish
- Portuguese
- Russian
- Arabic
Step 3: Generate Document
Click "Write a Patent Disclosure" button to begin generation.
Key Document Components
Background
Context and existing technology landscape
Detailed Description
Complete technical specification of invention
Technical Drawings
Diagrams and visual representations
Embodiments
Specific embodiments and variations
Patent Claims
Formal claims defining invention scope
Keywords & References
Strategic keywords and technical references
Export Options
- DOCX Format: Editable document for review and submission
- Markdown Format: Version control friendly format
Gatsbi Writer
Transform Your Existing Research into Professional Documents
Gatsbi Writer (introduced in version 2.0) specializes in converting your existing research, data, and findings into polished academic manuscripts or patent disclosures. Unlike Gatsbi Innovator which generates ideas from scratch, Gatsbi Writer works with your completed research.
Key Features
Research Integration
Import existing research via text or DOCX files
Smart Summarization
AI analyzes and structures your research content
Multiple Research Types
Supports methodological research, empirical research and case studies
Preserve Accuracy
Maintains your original data and findings
When to Use Gatsbi Writer
- You have completed research with results and data
- You need to transform rough notes into formal documents
- You want to repurpose research for different formats
- You need professional formatting and structure
Workflow
- Select "Gatsbi Writer" from the service menu
- Input your research (text or DOCX file)
- Review AI-generated summary
- Choose output format (paper, patent)
- Export in your preferred format
Learn More: See Write from Prior Research for detailed instructions.
Write from Research Notes or Existing Results
Transform your existing research into polished manuscripts or patent disclosures.
This feature is part of Gatsbi Writer - the specialized tool for converting existing research into professional documents.
Getting Started
Step 1: Select Gatsbi Writer
Choose "Gatsbi Writer" from the service menu.
Step 2: Provide Research Input
You can provide your research in two ways:
- Type or paste research description directly
- Attach a .DOCX document containing your research
Step 3: Automatic Research Type Recognition
After you input or attach your prior research materials, Gatsbi Writer automatically identifies the research type (methodological, empirical, or case study) based on your content. If the automatic recognition is incorrect, you can explicitly specify the research type by including it in your title input. For example: "On the effectiveness of online learning: A Case Study" clearly indicates a case study research.
Step 4: Research Summarization
Gatsbi Writer generates a comprehensive research summary focusing on:
- Methodology details
- Key results and findings
- Critical contributions
- Implications and applications
Minimum Input Requirements
If your prior research consists of a methodology study, i.e. you propose a novel method, model or process to solve a technical problem, the minimum input for Gatsbi Writer is a description of your methodology. You can also provide the experimental results if they are available. If your prior research involves experiments or statistics only, then it will be considered as an experimental study, in which case, the experimental results are required for Gatsbi Writer to work. Similarly, if you want to write a case study paper, you need at least provide either narrative or quantitative findings of the case study to Gatsbi Writer. If any of the required information above is missing, Gatsbi Writer will display a reminder message and prompt you to provide it.
| Study Type | Required Information |
|---|---|
| Methodology Study | Description of novel method or model |
| Experimental Study | Experimental results and data |
| Case Study | Narrative or quantitative findings |
Tip: Lengthy context may be compressed during the writing process. To ensure accuracy of critical information (such as experimental results), it's recommended to keep only core content (such as methodology, research findings, or results) in your input document while removing redundant sections like introduction and conclusion. Gatsbi will automatically reorganize and regenerate them.
Output Options
After summarization, you can generate:
- Full research manuscript
- Patent disclosure document (only for methodological research)
The Humanizer Plugin
Enhance AI-generated manuscripts with natural, human-like academic language.
Purpose: Make AI-written papers harder to detect while preserving technical accuracy and improving readability.
Key Benefits
- Reduces AI detection scores to nearly 0%
- Preserves technical accuracy and logical coherence
- Improves academic writing fluency
- Maintains original meaning and content
How to Use
Step 1: Generate Your Manuscript
First, create your manuscript using Gatsbi's standard tools.
Step 2: Access Humanizer
Click the "Humanizer" button near the export option.
Step 3: Processing
The plugin rewrites text paragraph-by-paragraph to:
- Vary sentence structure
- Add natural transitions
- Introduce stylistic variations
- Enhance readability
Credit System
Usage: 100 credits per 1,000 words processed
Credits can be tracked and purchased in-app through the credit management system.
Comparison
| Aspect | Original AI Text | After Humanizer |
|---|---|---|
| AI Detection Score | 80-95% | 0-5% |
| Readability | Good | Excellent |
| Academic Tone | Consistent | Natural variation |
| Technical Accuracy | 100% | 100% preserved |
Pro Tips
- Use for final manuscript polishing
- Helpful for journal submissions with AI-detection concerns
- Review output to ensure meaning preservation
The Plugin Credit System
Understanding and managing credits for advanced Gatsbi features.
What Are Plugin Credits?
Plugin credits are a separate resource system for accessing advanced add-on features like the Humanizer plugin. They:
- Enable premium plugin functionality
- Do not affect core tools (Gatsbi Writer, Gatsbi Innovator)
- Provide flexible usage options
Credit Types
Free Monthly Credits
Allocated each billing cycle. Note: Unused free credits do NOT roll over to the next month.
Purchased Credits
Buy additional credits anytime. These credits never expire and remain in your account.
How to Buy Credits
- Click the Balance area in the top right corner of the paper writing interface
- Click the "Buy More Credits" button
- Select credit package
- Complete secure payment
- Credits instantly added to account
Credit Consumption Order
The system automatically uses credits in this order:
- Free monthly credits - Used first
- Purchased credits - Used when free credits exhausted
Handling Insufficient Credits
Important: You can use the humanizer as long as your total credit balance is positive. If credits are insufficient to cover your usage, your balance will become negative. Next month's gift credits or any purchased credits will first be applied to compensate the negative balance.
Solutions:
- Wait for monthly credit refresh
- Purchase additional credits
Gatsbi Reviewer
AI-Powered Systematic Literature Reviews and Meta-Analyses
Gatsbi Reviewer automates the complex process of conducting systematic literature reviews and meta-analyses. It searches, screens, extracts, and synthesizes research literature following established academic standards like PRISMA guidelines.
Key Capabilities
Automated Search
Searches multiple academic databases simultaneously
AI Screening
Intelligently filters studies based on relevance
Data Extraction
Automatically extracts quantitative data for meta-analysis
Statistical Analysis
Generates forest plots, funnel plots, and effect sizes
Review Types Supported
| Type | Purpose | Output |
|---|---|---|
| Systematic Literature Review | Comprehensive narrative synthesis | Thematic analysis, quality assessment |
| Meta-Analysis | Statistical synthesis of quantitative studies | Effect sizes, heterogeneity analysis, plots |
PRISMA Compliance
All reviews follow PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines:
- Structured search strategy
- Clear inclusion/exclusion criteria
- PRISMA flow diagram
- Risk of bias assessment
- Transparent reporting
Learn More: See Systematic Literature Review and Meta-Analysis guides for step-by-step instructions.
Systematic Literature Review
Automatically generate comprehensive Systematic Literature Reviews with AI-powered analysis.
This feature is part of Gatsbi Reviewer - the specialized tool for evidence synthesis and literature analysis.
Getting Started
Select Gatsbi Reviewer from the service menu to access systematic review features.
4-Step Process
Step 1: Enter Research Topic
- Input your research topic or question in the search box
- Leave meta-analysis checkbox unchecked for systematic review only
- Optional: Set earliest publication year to filter recent studies
Step 2: Automated Search and Screening
The system automatically:
- Searches across major academic databases simultaneously
- Uses AI-powered screening to identify most relevant studies
- Filters based on relevance and methodological quality
- Identifies eligible studies based on your research question
Step 3: Review and Curate Studies
Interactive curation features:
- Generate structured outline of candidate studies
- Select/deselect studies for inclusion with checkboxes
- Upload custom PDFs for additional studies not found in search
Step 4: Write Manuscript
After curation, Gatsbi Reviewer immediately generates your systematic review manuscript with:
- Comprehensive narrative synthesis of findings
- Thematic analysis across all included studies
- Structured sections following PRISMA guidelines
- Discussion of implications and research gaps
- Properly formatted citations and references
Key Features
Automatic Citations
Properly formatted in-text citations and complete reference list
Structured Sections
PRISMA-compliant manuscript structure
Quality Assessment
Risk of bias evaluation and study quality ratings
Thematic Analysis
Comprehensive synthesis of qualitative findings
Sample Systematic Review Output
Meta-Analysis with Gatsbi Reviewer
Perform statistical synthesis of quantitative studies with automated data extraction and analysis.
This feature is part of Gatsbi Reviewer - specialized for quantitative evidence synthesis and statistical analysis.
Process for Meta-Analysis
Step 1: Enter Research Topic
- Input your research topic or question in the search box
- Check the meta-analysis checkbox to enable quantitative synthesis
- Optional: Set earliest publication year to filter recent studies
Step 2: Automated Search and Screening
The system automatically:
- Searches across major academic databases for quantitative studies
- Uses AI to identify studies with comparable outcome measures
- Filters studies suitable for statistical pooling
- Screens based on methodological quality and data availability
Step 3: Review and Curate Studies
Enhanced curation for meta-analysis:
- Generate structured outline showing candidate studies
- Review study designs and verify outcome measure compatibility
- Select/deselect studies with sufficient statistical data
- Upload custom PDFs for studies not found in automated search
Step 4: Data Extraction and Analysis
Automated extraction and analysis process (10-20 minutes):
- Automatically extracts effect sizes and confidence intervals
- Calculates pooled estimates using appropriate statistical models
- Assesses heterogeneity using I² statistics and Q-tests
- Performs subgroup and sensitivity analyses
- Generates publication bias assessments
Processing Time: Data extraction typically takes 10-20 minutes depending on the number of studies and complexity of data.
Step 5: Write a Paper Manuscript
After data extraction completes, click the "Write a Paper Manuscript" button to generate your comprehensive meta-analysis manuscript. The system will automatically integrate all statistical results, forest plots, funnel plots, and data tables into a properly formatted academic paper following journal guidelines.
Statistical Outputs
Forest Plots
Visual representation of effect sizes across studies
Funnel Plots
Assessment of publication bias
Heterogeneity Analysis
I² statistics and Q-test results
Data Export
Export analysis results before manuscript writing
Requirements for Meta-Analysis
- Recommended minimum of 10 studies with quantitative data
- Comparable outcome measures across studies
- Sufficient statistical information (means, SDs, CIs, effect sizes)
- Similar study designs or adjustable for heterogeneity
Tip: After data extraction completes, the system will display studies from which data was successfully extracted. At this point, while 10 studies are recommended for robust meta-analysis, if you have fewer studies you can either: (1) manually upload additional PDFs and re-run data extraction, or (2) use the "Force Write" option to proceed with your current study set when appropriate for your research context.
Sample Meta-Analysis Output
Payment Methods
Credit/Debit Card (via Stripe)
Subscription Model
Payments made through Stripe are set up as recurring subscriptions by default:
- Monthly Pro: Automatically renews every 30 days
- Yearly Pro: Automatically renews every 365 days
Supported Cards
We accept all major credit and debit cards through Stripe's secure payment processing.
WeChat Pay / 微信支付
One-Time Payment / 一次性付款: Each payment activates a 30-day usage period (non-recurring) / 每次支付激活30天使用期(非循环扣款)
Currency / 货币: Chinese Yuan (RMB) only / 仅支持人民币
Additional Fee / 额外费用: 13% VAT is added to the base price / 在原价基础上需额外支付13%增值税
Manage Subscription
Access your user dashboard on the website and select "Subscription" from the left menu to view your payment history and next billing date.
For Stripe Users
Click the "Subscription Management" button to access the Stripe portal where you can:
- Upgrade your current subscription plan
- Cancel your subscription (Note: Cancellation stops future charges at the next billing date. Your access remains active until then)
For WeChat Pay Users
WeChat Pay subscriptions are one-time purchases that expire automatically after the validity period with no auto-renewal.
Invoices
For Stripe Users
Stripe automatically generates and sends invoices and receipts to your registered email. To customize your invoice details:
- Preferred method: Enter your company name, VAT number, and other details directly during checkout
- Alternative: Contact Customer Support for manual invoice generation
For WeChat Pay Users / 微信支付用户
For Chinese VAT Invoice (Mainland China entities only) / 如需开具中国增值税发票(限中国大陆实体):
Please contact Customer Support and provide the following information / 请联系客服并提供以下信息:
Invoice Type / 发票类型
- VAT Special Invoice / 增值税专用发票
- VAT General Invoice / 增值税普通发票
⚠️ Note / 注意: Invoice item will be listed as "Software" and cannot be changed / 发票项目为"软件",不可更改
Required Information / 所需信息
For VAT General Invoice / 增值税普通发票:
- Company Name / 单位名称
- Taxpayer Identification Number / 纳税人识别号
For VAT Special Invoice / 增值税专用发票:
- Company Name / 单位名称
- Taxpayer Identification Number / 纳税人识别号
- Address / 地址
- Phone Number / 电话
- Bank Name and Account Number / 开户行及账号
ℹ️ Important / 重要说明
Stripe payment users cannot obtain Chinese VAT invoices. However, according to Chinese tax policy, the receipts automatically issued by Stripe can be directly used for reimbursement as proof of purchasing overseas software services.
Stripe付款用户无法开具中国增值税发票,但根据中国税务政策,Stripe自动发送的收据可作为购买境外软件服务的凭据直接用于报销。
详见:《企业所得税税前扣除凭证管理办法》第十一条
https://www.gov.cn/zhengce/zhengceku/2019-10/23/content_5444239.htm
Customer Support
Get help quickly through multiple support channels.
Contact Methods
Email Support
Send an email directly to helpdesk@gatsbi.com for assistance with any questions or issues.
Support Ticket System
Log in to your user space on the website and submit a helpdesk ticket for customer support requests.
Reporting Technical Issues
Best Practices for Technical Support:
- Include screenshots of any error messages or issues
- Describe the steps that led to the problem
- Specify which Gatsbi tool you were using
- Include your browser and operating system information
Tip: When encountering technical problems, please attach screenshots to help our support team understand and troubleshoot the issue more quickly.
Response Times
- Typically respond within 24 hours
- During holidays, response time may extend to 48 hours
- Both email and ticket systems follow the same response timeline
Self-Service Resources
- Check this user guide for detailed instructions
- Visit our FAQ section on the website
- Watch tutorial videos on the website
We're Here to Help! Our support team is committed to ensuring you have a smooth experience with all Gatsbi tools. Don't hesitate to reach out with any questions or concerns.