---
title: AI assistants and agents
slug: q9ad0472
canonical_url: https://docs.coveo.com/en/q9ad0472/
collection: integrations
source_format: adoc
---
# AI assistants and agents

Coveo integrates with LLM apps and agent platforms to ground responses in relevant enterprise content.
By connecting these experiences to Coveo, organizations can give AI systems governed access to enterprise knowledge while keeping the experience in the platform their teams use.

This page is for developers and system administrators evaluating or building a Coveo integration with an LLM app or agent platform.

## What Coveo powers in these platforms

LLM apps and agent platforms help organizations understand requests, find information, choose actions, and complete tasks.
Coveo adds retrieval capabilities that help these systems find relevant enterprise content or process a request before generating a response or taking an action.
The AI system uses the returned information according to the permissions configured for the [Coveo organization](https://docs.coveo.com/en/185.md).

## Use cases

* Ground LLM app and agent responses in passages from content [indexed](https://docs.coveo.com/en/204.md) in Coveo.

* Help users find and summarize information from large and complex enterprise content sources.

* Extend an AI experience with Coveo retrieval while keeping the experience in the platform users already use.

* Classify support cases in real time and use the result in an agent workflow.

## Typical integration pattern

The integration pattern depends on the capabilities of the target platform.

For an LLM app or agent platform that connects through an MCP client, the MCP client sends requests to the Coveo MCP Server.
The server returns the results of the requested Coveo tool, and the AI system uses those results to generate a context-aware response.
The following diagram shows this representative MCP pattern; packaged integrations and gateway-mediated connections can use different components.

![MCP-based integration between Coveo Platform and an LLM app](https://docs.coveo.com/en/assets/images/integrations/llm-app-category-integration-diagram.svg)

| 1 | The MCP client in the LLM app sends requests to the Coveo MCP Server. The server returns the results of the requested Coveo tool to the client, and the LLM app uses those results to generate a response. |
|---|---|

Refer to [product-specific articles](#product-specific-articles) when available.

## Product-specific articles

The following LLM apps, agent platforms, and agent frameworks have dedicated Coveo integration articles.

- [Agentforce](https://docs.coveo.com/en/q93b0002.md)