tydids

Dashboard

Welcome,

-

Collections

-

Documents

-

Searches

-

Uploads

Quick Search

Recent Collections

Recent Activity

Session Details

Session description

unknown

Research Progress

0% complete

Execution Log

Waiting for execution logs...

Research Results

No results generated yet.

Collection Details

-

Documents

-

Chunks

-

Size

-

Last Updated

AI Assistant for "Collection"

Ask questions about your documents and get intelligent responses with source attribution.

Start a conversation about your documents!

Try asking:

• "What are the main topics in these documents?"
• "Summarize the key findings"
• "What does the document say about [topic]?"

Responses are generated using Google Gemini AI based on your documents

Chat Settings

Max Sources Number of document chunks to use for context

Custom Instructions Leave empty for default behavior

Quick Actions

Smart Context Generator

Create intelligent context for AI/LLM applications using semantic search enhanced with cluster information. Perfect for optimizing context windows while maintaining relevance and diversity.

Context Configuration

Search Query

Max Context Size (characters)

Max Chunks

Diversity Weight

Cluster Weight

Include cluster metadata in context

Collection Info

Loading capabilities...

Simple File Upload API
Upload files to this collection using a simple multipart/form-data request with just a "file" field. No additional metadata required - the backend handles everything automatically.

Upload files to this collection programmatically
Use these code snippets to integrate file uploads into your applications. Replace the token and collection ID with your actual values.

Upload Endpoint

Loading...

AI Chat Endpoint

Loading...

Collection ID

Loading...

Authentication Token: Loading...

cURL Command

Loading...

Node.js with Fetch API

Loading...

JavaScript with Axios

Loading...

Python with Requests

Loading...

VSI Vector Store API

Comprehensive API suite for document management, AI interactions, and intelligent content organization. Upload documents, ask AI questions, generate smart context, and organize content with clustering.

Document Upload

Simple file upload with automatic processing:

PDF, Word, Text, Markdown, Images
Automatic text extraction
Vector embeddings generation
Immediate search availability

AI Chat

Intelligent document Q&A:

Semantic search across documents
Context-aware responses
Multiple AI model support
Source attribution

Smart Context

Intelligent context generation:

Semantic search & clustering
Diversity optimization
Token-aware sizing
Multiple strategies

Document Clustering

Automatic content organization:

K-means, hierarchical, DBSCAN
Theme discovery
Outlier detection
Performance analytics

Quick Start - File Upload:

Upload: Send a POST request with a "file" field to the upload endpoint
Processing: The backend automatically extracts text, creates embeddings, and stores in the vector database
Ready: Documents are immediately available for AI chat, smart context, and clustering

Supported File Types:

PDF - Portable Document Format
DOCX - Microsoft Word Documents
DOC - Legacy Word Documents
TXT - Plain Text Files
MD - Markdown Files
Images - JPG, JPEG, PNG, GIF, BMP, WebP (with AI analysis)

Request Format:

Content-Type: multipart/form-data
Authorization: Bearer YOUR_JWT_TOKEN

Form Data:
- file: [binary file data]

Response Format:

{
  "success": true,
  "message": "Document processed successfully",
  "document": {
    "id": 123,
    "filename": "document.pdf",
    "file_type": "pdf",
    "collection_id": 15
  },
  "chunksStored": 25,
  "totalChunks": 25,
  "processingTime": 3500
}

Pro Tip: After uploading, wait 2-5 seconds before using AI features to ensure all embeddings are processed.

Smart Context API

Generate intelligent context from your documents using semantic search and clustering. This API creates optimized context for LLM applications by finding relevant content and ensuring diversity.

Endpoints:

POST /api/collections/:id/smart-context - Generate smart context
POST /api/collections/:id/smart-context/preview - Preview context generation
GET /api/collections/:id/smart-context/capabilities - Get collection capabilities

cURL Example:

Generate smart context

Loading...

JavaScript Example:

Using fetch API

Loading...

Python Example:

Using requests library

Loading...

Request Parameters:

query - Search query to find relevant content
maxTokens - Maximum tokens in generated context (default: 4000)
diversityThreshold - Diversity threshold 0-1 (default: 0.7)
includeMetadata - Include source metadata (default: true)
strategy - 'balanced', 'diversity', or 'relevance' (default: 'balanced')

Strategies:
• Balanced: Optimal mix of relevance and diversity
• Diversity: Prioritizes content variety
• Relevance: Focuses on most relevant content

Clustering API

Organize your documents into semantic clusters for better navigation and understanding. The clustering API groups similar content together using advanced machine learning algorithms.

Endpoints:

POST /api/collections/:id/clusters - Create document clusters
GET /api/collections/:id/clusters - Get existing clusters
GET /api/collections/:id/clusters/:clusterId/documents - Get cluster documents

cURL Example:

Create clusters

Loading...

JavaScript Example:

Fetch clusters

Loading...

Python Example:

Using requests library

Loading...

Clustering Parameters:

numClusters - Number of clusters to create (default: auto-detect)
algorithm - 'kmeans', 'hierarchical', or 'dbscan' (default: 'kmeans')
includeOutliers - Include outlier documents (default: true)
minClusterSize - Minimum documents per cluster (default: 2)

Response Format:

{
  "clusters": [
    {
      "id": 1,
      "name": "Machine Learning",
      "description": "Documents about ML algorithms and techniques",
      "size": 15,
      "centroid": [0.1, 0.2, ...],
      "documents": ["doc1", "doc2", ...]
    }
  ],
  "analytics": {
    "totalClusters": 5,
    "silhouetteScore": 0.65,
    "processingTime": 2300
  }
}

Use Cases:
• Organize large document collections
• Discover content themes and patterns
• Improve search and navigation
• Generate topic summaries

⚠️ Important Notes

Simple Upload: Just send the file - no additional fields required
Authentication: Your current JWT token is included in all examples above
File Processing: Allow 2-10 seconds for document processing before using AI features
Rate Limiting: API calls are subject to your account tier limits
AI Features: Upload, chat, smart context, and clustering all use semantic search
Smart Context: Optimizes content selection for LLM applications with diversity control
Clustering: Organizes documents into semantic groups for better navigation
Token Security: Keep your token secure and do not share these examples with others

Authentication

All API requests require a valid JWT token in the Authorization header. You can get your token by logging into the web interface and checking your browser's localStorage.

Supported Formats

PDF, Word Documents (DOCX), Text files (TXT), Markdown (MD), and Images (JPG, JPEG, PNG, GIF, BMP, WebP) for AI analysis.

How it works:
Drag the bookmarklet below to your browser's bookmark bar. When you click it on any webpage, it will:

Selected text: Upload the selected text as a new document to this collection
No selection: Take a screenshot of the visible area and upload it as an image

Drag this to your bookmark bar:

📋 Quick Upload to Collection

← Drag this button to your bookmark bar

Or right-click and "Add to bookmarks" / "Bookmark this link"

Text Selection

Select any text on a webpage and click the bookmarklet. The selected text will be uploaded as a new document with the page title and URL as metadata.

Screenshot Capture

If no text is selected, the bookmarklet will capture a screenshot of the visible area and upload it as an image for AI analysis.

Security & Privacy

Authentication: Your personal JWT token is embedded in the bookmarklet
Collection-specific: Only uploads to this specific collection
Privacy: No data is sent to third parties - only to your tydids instance
HTTPS required: Screenshots only work on HTTPS websites due to browser security
Permissions: Browser may ask for screenshot permission on first use

Bookmarklet Code

Raw JavaScript (for manual setup):