# Nomic Atlas

> Nomic Atlas is a platform and Python client for embedding, visualizing, searching, and sharing large unstructured datasets, with an Agent API for AEC document workflows.

Nomic Atlas is a product from Nomic AI for working with unstructured data. The nomic Python client lets developers create maps from text, images, or raw embeddings, then explore, search, and share them in a web browser. The current Nomic site also presents an Agent API and platform for parsing, extracting, and embedding drawings, specs, and other project files in architecture, engineering and construction (AEC).

## What It Is

Atlas is an unstructured data interaction platform. According to the repository README, it supports datasets from hundreds to tens of millions of points across text, image, audio, and video. The Python package (`pip install nomic`) is authenticated with `nomic login` and a token. A call such as `atlas.map_data(embeddings=...)` creates a dataset and map.

## How the Python Client Works

Atlas stores, manages, and generates embeddings for your data. From a map you can download latent (high-dimensional) embeddings and the 2D projected embeddings used for display. It automatically groups data into topics, exposed programmatically as a pandas DataFrame at multiple topic depths. Vector search finds nearest neighbors for a given datapoint, and the README lists clustering, tagging, cleaning, and deduplication of text, images, video, and audio.

## Agent API for AEC Files

The Agent API exposes Parse, Extract, Embed, and Datasets endpoints. Parse turns large drawing sets, specs, and complex PDFs into structured representations with layout, OCR, tables, and visual context. Extract produces structured outputs from document collections using schemas and retrieval. Embed creates multimodal embeddings for retrieval. Datasets stores and governs project data collections. Nomic states these are the same models behind its Nomic Platform.

## Deployment and Integrations

Nomic says the API runs in Nomic-managed, per-tenant AWS environments with hosting in the EU, AU, US, or UK. It lists connectors for SharePoint, Egnyte, Autodesk Construction Cloud, ProjectWise, Trimble ProjectSight, Fieldwire, Google Drive, Box, Dropbox, Microsoft Teams, Outlook, and Gmail. It also lists MCP server connections.

## Features
- Generate, store, and retrieve embeddings for unstructured data
- Interactive browser-based data maps
- Automatic topic modeling and clustering
- Semantic and vector nearest-neighbor search
- Tagging, cleaning, and deduplication of text, images, video, and audio
- Python client with programmatic access to embeddings and topics
- Agent API with Parse, Extract, Embed, and Datasets endpoints
- Multimodal search across drawings, specs, tables, and scanned PDFs
- Multi-region hosting in EU, AU, US, and UK

## Integrations
SharePoint, Egnyte, Autodesk Construction Cloud, ProjectWise (Bentley), Trimble ProjectSight, Fieldwire, Google Drive, Box, Dropbox, Microsoft Teams, Outlook, Gmail, MCP servers

## Platforms
WEB, API, DEVELOPER_SDK, CLI

## Pricing
Freemium — Free tier available with paid upgrades

## Version
v3.9.0

## Links
- Website: https://atlas.nomic.ai
- Documentation: https://docs.nomic.ai/
- Repository: https://github.com/nomic-ai/nomic
- EveryDev.ai: https://www.everydev.ai/tools/nomic-atlas