Chat

Build Conversational AI into your Apps with RAG

Chat with retrieval-augmented generation (RAG) integrates inputs, sources, and models to build more powerful product experiences. It’s all powered by Command.

Get your API Key

chat with our model

Our Customers

Retrieve information with connectors

VIEW THE DOCS

Enterprise Datastores

Point the model at internal datastores, allowing it to seamlessly (and securely) cite your proprietary data.

The Internet

Point the model at the web, allowing it to generate responses grounded in real-time information.

Specific Documents

Provide the model with manually selected documents, enabling grounded Q&A on particular content.

CONNECT TO DATA SOURCES LIKE:

What’s possible when Chat and RAG come together?

Conversational Knowledge Assistants

Create assistants to automate tasks and give grounded answers to your questions through natural interactions and connections to data.

Customer Support

Provide immediate answers from knowledge bases to support agents and engage customers with prompt resolutions for complex escalations.

Learning apps

Deliver dynamic and personalized learning experiences with interactive lessons, quizzes, and feedback based on evolving user profiles.

Why Chat with RAG?

Conversation  is the new interface

Chat understands the intent behind messages, remembers conversation history, and responds intelligently through multi-turn conversations. Chat responses are powered by Cohere's Command model.

Reduce hallucinations with grounding, citations

Reduce hallucinations and create trust between generated responses and users with citations to understand where responses are coming from. Command is trained to answer questions from additional sources.

Keep your data private

When privately deployed, the training data, input prompts, and output responses stay private and don't leave your secure environment.

Simple APIs, powerful results

No matter your level of experience with ML/AI, Cohere’s Command model makes it easy to build chat interfaces in your applications.

Get started

API Reference

1import cohere 
2co = cohere.Client('MLZXavfC2EpNaW3dYRG5KwWPcMIvBUyabF1DPBgw') # This is your trial API key
3response = co.chat( 
4  message='<YOUR MESSAGE HERE>',
5  prompt_truncation='auto',
6  connectors=[{"id": "web-search"}]
7) 
8print(response)