Nomic Atlas

Overview

Atlas is an AI-ready data layer and visualization platform for unstructured data, analytics, and AI workflows. It enables visualization of massive datasets of embeddings directly in web browsers with second-range latencies.

Key Features

Visualization Capabilities

• Massive Scale: Upload and visualize millions of embeddings
• Interactive Maps: 2D coordinate representations with automatic topic clustering
• Dimensionality Reduction: Uses t-SNE, UMAP, or PCA to reduce high-dimensional vectors to 2D/3D
• Semantic Clustering: Automatically assigns colors based on semantic similarity
• Browser-Based: No local installation required, runs entirely in browser
• Jupyter Integration: Interactive notebooks support

Embedding Support

• Text Embeddings: State-of-the-art nomic-embed-text-v1.5 model
• Image Embeddings: nomic-embed-vision-v1.5 for visual data
• Multimodal: Handle text, PDFs, images, and mixed embeddings
• Custom Embeddings: Upload embeddings from any source

How It Works

1. Upload Data: Import embeddings from various sources
2. Automatic Projection: Atlas generates 2D projections using dimensionality reduction
3. Interactive Exploration: Each dot represents a data point, colored by semantic similarity
4. Topic Discovery: Automatically identify clusters and themes

Use Cases

• Semantic Search: Visualize search result quality and coverage
• RAG Applications: Analyze retrieval patterns and document coverage
• Clustering Analysis: Discover natural groupings in unstructured data
• Model Debugging: Understand embedding model behavior
• Data Quality: Identify outliers and data issues
• Vector Database Visualization: Inspect contents of Pinecone, Weaviate, and other vector DBs

Integration

Vector Database Support

• Weaviate
• Pinecone
• Custom vector stores
• Direct embedding uploads

APIs and SDKs

• Python SDK (nomic)
• Web API for programmatic access
• Jupyter notebook widgets
• OpenAI Cookbook examples

Platform Capabilities

Data Operations

• Upload millions of data points
• Real-time collaborative editing
• Embedding generation at scale
• Retrieval and semantic search
• Data versioning and lineage

Analytics Features

• Topic modeling
• Semantic similarity analysis
• Cluster statistics
• Embedding quality metrics
• Data drift detection

Performance

• Latency: Second-range response times for massive datasets
• Scale: Handles millions of embeddings
• Interactive: Real-time exploration without lag
• Efficient: Optimized projection algorithms

Pricing

Atlas offers free tier with limitations and paid plans for larger datasets and teams. Details available on Nomic AI website.