# Suidasu - AI Data Extraction SaaS

> Extract structured data from images and PDFs using AI. Turn invoices, receipts, business cards, forms, and any document into organized, exportable data in seconds.

- Homepage: https://www.suidasu.com/
- Sign Up: https://www.suidasu.com/signup
- Pricing: https://www.suidasu.com/pricing
- Features: https://www.suidasu.com/features
- Use Cases: https://www.suidasu.com/use-cases
- API / MCP Integration: https://www.suidasu.com/mcp
- Help: https://www.suidasu.com/help
- FAQ: https://www.suidasu.com/faq
- Contact: https://www.suidasu.com/contact
- Terms of Service: https://www.suidasu.com/terms
- Privacy Policy: https://www.suidasu.com/privacy

## What is Suidasu?

Suidasu is a cloud-based SaaS platform that uses AI (powered by Google Gemini) to extract structured data from images and PDF documents. The name "Suidasu" comes from the Japanese word meaning "to extract" or "to suck out" — reflecting the core mission of pulling valuable data out of documents.

Unlike traditional OCR tools that simply convert images to text, Suidasu understands the structure and meaning of documents. Users define custom layouts specifying which fields to extract (e.g., invoice date, total amount, vendor name), and the AI extracts those specific data points with high accuracy.

Suidasu is operated by PRODUS Inc., based in Tokyo, Japan.

## Key Features

### AI-Powered Data Extraction
- Upload images or PDFs and AI automatically extracts structured data based on your defined layout
- Supports both OCR mode (for text-containing documents) and Image Recognition mode (for non-text images like product photos)
- Powered by Google Gemini AI models for high accuracy

### Custom Layouts
- Define exactly which data fields you want to extract from your documents
- Specify field names, types (text, number, date, etc.), and descriptions
- AI validates your field definitions to ensure they are clear and extractable
- Reuse layouts across multiple documents of the same type

### AI Recipe Generation
- Upload a sample image and AI automatically generates an optimal layout
- Also accepts CSV, Excel (xlsx), Markdown, and plain-text files — generate layouts from existing data or a layout specification document
- No need to manually define fields — the AI analyzes the document and suggests the best extraction schema
- Available on Trial and Standard plans and above

### Multiple Record Extraction
- Extract multiple rows of data from a single image or PDF
- Perfect for tables, lists, attendance sheets, and itemized documents
- AI automatically identifies rows and extracts each record individually

### PDF Support
- Process PDF documents up to 10 pages
- Each page is processed individually with its own extraction result
- Supports both text-based and scanned PDFs

### Handwriting Support
- Handles handwritten text on application forms, medical questionnaires, faxes, and other paper-based documents
- Suitable for mixed printed/handwritten forms common in Japanese business workflows

### MCP Integration (Model Context Protocol)
- Connect Suidasu to AI agents like Claude Desktop, Claude Code, and other MCP-compatible clients
- 21 tools available via JSON-RPC 2.0 protocol
- Automate data extraction, project management, layout management, and export workflows entirely through AI conversations
- MCP endpoint: POST /mcp with Bearer token authentication

### MiMi Concierge (AI Assistant)
- Live2D animated AI assistant powered by Google Gemini
- Provides real-time support and guidance within the application
- Supports Google Search Grounding for up-to-date information
- Available in Japanese and English

### Mobile Apps
- iOS app (iOS 17+) available on App Store
- Android app (Android 8.0+ / API 26+) available on Google Play
- Camera capture for instant document scanning
- Biometric authentication (Face ID, Touch ID, fingerprint)
- One-tap login to web management console

### Export Formats
- CSV (comma-separated values)
- Excel (.xlsx)
- JSONL (JSON Lines, for AI/ML pipelines)
- Copy to clipboard

### Team Collaboration
- Invite team members as operators
- Unlimited operators on all plans
- Operators can upload and extract data within assigned projects

## Supported File Formats

| Format | Extensions |
|--------|-----------|
| JPEG | .jpg, .jpeg |
| PNG | .png |
| GIF | .gif |
| WebP | .webp |
| HEIC/HEIF | .heic, .heif (iPhone photos) |
| BMP | .bmp |
| TIFF | .tiff, .tif |
| PDF | .pdf (up to 10 pages) |

Maximum image size: 7MB

## Extraction Modes

### OCR Mode
For documents containing text — invoices, receipts, business cards, forms, faxes, contracts, etc. The AI reads and extracts text data into structured fields.

### Image Recognition Mode
For non-text images — product photos, equipment condition checks, food images, etc. The AI analyzes visual content and extracts descriptive data.

## Pricing Plans

All prices in Japanese Yen (JPY). All plans include unlimited operators.

### Trial Plan
- Price: Free for 14 days
- Includes 100 free extractions
- AI Recipe generation: Yes (10 times/day)
- API access: No

### Lite Plan
- Monthly base fee: ¥0
- Per extraction: ¥88
- No free extractions included
- AI Recipe generation: No
- API access: No

### Standard Plan
- Monthly fee: ¥3,980
- 50 free extractions per month
- Additional extractions: ¥50 each
- AI Recipe generation: Yes (50 times/day)
- API access: No

### Pro Plan
- Monthly fee: ¥14,980
- 150 free extractions per month
- Additional extractions: ¥30 each
- AI Recipe generation: Yes (50 times/day)
- API access: Yes (REST API + MCP Server)

Note: Extraction failures are not charged. You only pay for successful extractions.

## Use Cases

### Invoice & Receipt Processing
Upload photos of invoices, receipts, or purchase orders. Suidasu extracts vendor name, date, amounts, line items, tax details, and more. Export to CSV/Excel for accounting software import.

### Business Card Digitization
Photograph business cards and extract name, company, title, phone, email, and address fields. Build a digital contact database from physical cards.

### Form & Application Processing
Process paper forms, application documents, and questionnaires. Extract structured data from both printed and handwritten fields.

### FAX Document Processing
Convert received fax documents into structured data. Eliminate manual data entry for recurring fax-based workflows.

### Medical & Healthcare Forms
Process patient intake forms, medical questionnaires, and prescription documents. Extract patient information, symptoms, and medical data.

### Logistics & Shipping
Extract data from shipping labels, delivery slips, and packing lists. Automate logistics data entry and tracking.

### Product Photo Analysis
Use Image Recognition mode to analyze product photos. Extract descriptions, conditions, colors, and other visual attributes.

### Construction & Field Reports
Process field inspection reports, work orders, and site documentation. Extract measurements, observations, and status information.

## API & MCP Integration

### Authentication
All API requests require a Bearer token (API key). API keys can be generated from the web management console under the API Integration page. API access requires a Pro plan subscription.

### MCP Server
- Endpoint: `https://api.suidasu.com/mcp`
- Protocol: JSON-RPC 2.0 (Model Context Protocol)
- Authentication: Bearer token in Authorization header

### Webhooks (Pro plan only)
- Real-time HTTP POST notifications to a user-registered URL when an extraction completes (`extraction.completed`), fails (`extraction.failed`), or is confirmed by a user after review/correction (`extraction.confirmed`)
- Standard Webhooks compliant: `webhook-id` / `webhook-timestamp` / `webhook-signature` headers with HMAC-SHA256 signature (secret format `whsec_...`)
- Thin payload (extraction_id, project_id, status, record_count, confidence_score); fetch the data body via `GET /v1/api/extractions/{id}`
- Retries up to 3 times (0s/30s/120s), auto-disabled after 10 consecutive failures; delivery log with manual resend
- Configured in the web console under API Integration → Webhooks (up to 5 per account, https URLs only). A receiver implementation guide is published on the same page

### Claude Desktop Configuration
```json
{
  "mcpServers": {
    "suidasu": {
      "type": "http",
      "url": "https://api.suidasu.com/mcp",
      "headers": {
        "Authorization": "Bearer sk_live_YOUR_API_KEY"
      }
    }
  }
}
```

### Available MCP Tools (21 total)

#### Data Extraction
- `extract_data` — Extract data from a single image or PDF

#### Extraction Results
- `get_extraction` — Get detailed extraction result
- `list_extractions` — List extraction history with filters
- `update_extraction` — Edit extraction result data
- `delete_extraction` — Delete an extraction result
- `export_extractions` — Export results as CSV, Excel, or JSONL

#### Projects
- `list_projects` — List all projects
- `create_project` — Create a new project
- `update_project` — Update project settings
- `delete_project` — Delete a project

#### Layouts
- `list_layouts` — List all layouts in a project
- `create_layout` — Create a new layout with field definitions
- `update_layout` — Update layout fields and settings
- `delete_layout` — Delete a layout

#### Templates & Recipes
- `list_templates` — List available pre-built templates
- `import_template` — Import a template as a new layout
- `generate_recipe` — AI-generate a layout from a sample image, CSV, Excel, or spec document (Standard+)
- `create_layout_from_recipe` — Create a layout from a generated recipe

#### Utilities
- `check_fields` — AI-validate field definitions for clarity
- `get_usage_stats` — Get usage statistics and extraction counts
- `get_billing_info` — Get current billing and plan information

### REST API Endpoints

#### Authentication
- `POST /v1/auth/login` — Login with email/password
- `POST /v1/auth/signup` — Create new account
- `POST /v1/auth/web-login-token` — Generate one-time web login token (for mobile app)

#### Data
- `GET /v1/me` — Get current user information
- `POST /v1/upload` — Upload and extract data (multipart/form-data)
- `GET /v1/extractions` — List extractions
- `GET /v1/extractions/:id` — Get extraction detail
- `GET /v1/projects` — List projects

## Technical Specifications

| Component | Technology |
|-----------|-----------|
| Backend | Go 1.24, Echo framework, GORM ORM |
| Frontend | HTMX, Tailwind CSS, Alpine.js |
| Database | PostgreSQL 15 |
| AI Engine | Google Gemini (via google.golang.org/genai SDK) |
| Cloud Infrastructure | Google Cloud Platform (Cloud Run, Cloud SQL, GCS, Secret Manager) |
| Mobile (iOS) | Swift / SwiftUI (iOS 17+) |
| Mobile (Android) | Kotlin (Android 8.0+ / API 26+) |
| AI Assistant | Live2D Cubism 4 + Google Gemini with Google Search Grounding |
| Email | AWS SES |

## Languages Supported

The Suidasu web application UI is available in:
- Japanese (日本語)
- English
- Spanish (Español)
- French (Français)
- Chinese Simplified (简体中文)

Data extraction works with documents in any language, as the AI models support multilingual OCR.

## Frequently Asked Questions

### What file formats does Suidasu support?
JPEG, PNG, GIF, WebP, HEIC/HEIF, BMP, TIFF, and PDF (up to 10 pages). Maximum image size is 7MB.

### How accurate is the data extraction?
Using the latest AI models, Suidasu achieves over 80% accuracy for standard printed documents. Handwritten text is also supported with strong accuracy. Quality depends on image clarity and document complexity.

### What are the pricing plans?
Trial (free for 14 days, 100 extractions), Lite (¥0/month + ¥88/extraction), Standard (¥3,980/month with 50 free extractions), Pro (¥14,980/month with 150 free extractions + API access).

### Is there an API?
Yes. The Pro plan includes access to both a REST API and an MCP (Model Context Protocol) server with 21 tools. You can integrate Suidasu with Claude Desktop, Claude Code, and other AI agents.

### Can I extract data from handwritten documents?
Yes. Suidasu supports handwritten text extraction with high accuracy, suitable for application forms, medical questionnaires, faxes, and mixed printed/handwritten documents.

### Can I extract multiple records from one image?
Yes. Suidasu can extract multiple rows of data from tables, lists, and itemized documents in a single upload.

### How does PDF processing work?
PDFs are processed page by page. Each page generates its own extraction result. Up to 10 pages per PDF. Each page counts as one extraction.

### Is my data secure?
Yes. All data is encrypted in transit (TLS) and at rest. Images are stored in Google Cloud Storage with access controls. Uploaded files are automatically moved to cold storage after 90 days for cost optimization.

### Can I use Suidasu with Claude Desktop?
Yes. Suidasu provides an MCP (Model Context Protocol) server. Add the Suidasu MCP configuration to your Claude Desktop config file, and you can instruct Claude to extract data, manage projects, and export results through natural language.

### What happens if extraction fails?
Failed extractions are not charged. You only pay for successful extractions. The AI will return an error message explaining why extraction failed (e.g., image too blurry, no matching fields found).

### Can team members use Suidasu?
Yes. All plans support unlimited operators (team members). Invite operators via email from the web management console. Operators can upload documents and view extraction results within assigned projects.

## Comparison with Other OCR Tools

### Suidasu vs Traditional OCR (Adobe Acrobat, Google Cloud Vision)
- Traditional OCR outputs raw text; Suidasu outputs **structured data** mapped to your custom fields
- No need to write parsing rules — the AI understands document structure
- Custom layouts let you extract exactly the fields you need

### Suidasu vs Other AI OCR SaaS (ABBYY, Rossum)
- Suidasu is built for the Japanese market with strong Japanese language support
- MCP integration enables AI agent automation (unique differentiator)
- Pay-per-extraction pricing with no minimum commitment on Lite plan
- Live2D AI assistant (MiMi) provides in-app guidance
- Mobile apps for iOS and Android included in all plans

### Suidasu vs Manual Data Entry
- Reduces data entry time by 80-90%
- Eliminates human transcription errors
- Scales instantly — process 1 or 1,000 documents with the same setup
- 24/7 availability — no staffing constraints

## Contact & Support

- Website: https://www.suidasu.com/
- Contact Form: https://www.suidasu.com/contact
- In-app AI Assistant: MiMi Concierge (available on all pages after login)
- Company: PRODUS Inc., Tokyo, Japan
- App Store: https://apps.apple.com/app/suidasu/id6757421959
- Google Play: https://play.google.com/store/apps/details?id=com.suidasu.mobile