Sold by: Coveo
Deployed on AWS
Power your generative and agentic applications with Coveo's intelligent Passage Retrieval API. An AWS-ready, developer-first solution for grounding AI assistants with enterprise-verified knowledge. Built to integrate with Amazon Bedrock and Q Business, Passage Retrieval is your composable retrieval layer for high-performance, secure, and scalable LLM experiences.
Overview
The Coveo MCP Server is a hosted Model Context Protocol endpoint that turns your Coveo index into retrieval tools any MCP-compatible agent can call. It is a remote server over streamable HTTP, authenticated with OAuth or an API key: point an agent at the endpoint and it can search, retrieve and cite your enterprise content, with no connectors to build, no vector database to operate and no second copy of your data to keep in sync.
FOUR TOOLS OVER ONE INDEX
Search returns ranked results from across everything Coveo indexes, using Coveo relevance and semantic models rather than keyword matching.
Passage Retrieval returns the ranked passages most relevant to a question, with source references, for grounding a model's response.
Answer generates a grounded answer with citations using Coveo Relevance Generative Answering.
Fetch retrieves a specific item by its unique identifier, for when an agent needs the full document rather than an excerpt.
Administrators configure each tool in the Coveo Administration Console, setting the name and the description a model reads when deciding whether to call it. Retrieval behavior is therefore tuned per agent, in configuration, rather than hard-coded in the application.
BUILT FOR AWS AGENTIC STACKS
Amazon Bedrock AgentCore, Amazon Quick Suite and Amazon Connect are tested MCP clients and can be configured with either an API key or OAuth. Coveo holds the AWS Generative AI Competency, participates in AWS ISV Accelerate, and was a launch partner for the AWS Marketplace AI Agents and Tools category.
PERMISSION-AWARE BY DEFAULT
Coveo imports each content source's native permission model when it crawls. Using an early-binding approach, content a user is not entitled to is excluded before the query runs rather than filtered afterward. Use OAuth when agents act on behalf of authenticated users and secured content is in scope; use an anonymous API key for public content. Each MCP configuration receives its own search hub, so every agent's relevancy is tailored and retrieval traffic is separately reportable in Coveo analytics.
NOT ONLY AWS
Tested clients also include ChatGPT Enterprise, where Coveo is listed in the Apps and Connectors directory, along with Claude, Microsoft 365 Copilot, Microsoft Copilot Studio, Gemini Enterprise, Agentforce, Workato, Figma Make, Cursor and Visual Studio Code. One configuration serves all of them.
SECURITY AND COMPLIANCE
ISO/IEC 27001, 27017, 27018 and 27701 certified. SOC 2 Type II examined annually. HIPAA-compliant deployment option available. Encryption in transit and at rest, single sign-on, and selectable data residency across AWS regions in North America, Europe and Australia.
The MCP Server is part of the Coveo AI-Relevance Platform. Coveo serves more than 700 brands and is a Leader in the 2026 Gartner Magic Quadrant for Search and Product Discovery. Purchasing through AWS Marketplace consolidates Coveo on your AWS invoice. Contact Coveo for a private offer sized to your query volume.
Highlights
- Four retrieval tools, one endpoint: Search for ranked results, Fetch for a specific item by ID, Passage Retrieval for citable passages that ground a model's response, and Answer for a generated answer with references. Administrators name each tool and write the description the model reads before calling it, so retrieval behavior is tuned per agent in configuration instead of hard-coded in your application.
- Permission-aware retrieval, not a second copy of your data: Coveo imports each source's native permissions at crawl time, so unauthorized content is excluded before the query runs rather than filtered afterward. Index Salesforce, ServiceNow, SharePoint, Confluence, Adobe, Amazon S3 and more in place, and use OAuth so an agent retrieves only what the signed-in user is entitled to see.
- Drops into your AWS agentic stack: One remote endpoint over streamable HTTP, with Amazon Bedrock Agent/AgentCore, Amazon Quick Suite and Amazon Connect as tested MCP clients supporting API key or OAuth. The same endpoint also serves ChatGPT Enterprise, Claude, Microsoft 365 Copilot and Gemini Enterprise.
Details
Sold by
Delivery method
Type
Deployed on AWS
New
Introducing multi-product solutions
You can now purchase comprehensive solutions tailored to use cases and industries.
Features and programs
Financing for AWS Marketplace purchases
AWS Marketplace now accepts line of credit payments through the PNC Vendor Finance program. This program is available to select AWS customers in the US, excluding NV, NC, ND, TN, & VT.
Pricing
Pricing is based on the duration and terms of your contract with the vendor. This entitles you to a specified quantity of use for the contract duration. If you choose not to renew or replace your contract before it ends, access to these entitlements will expire.
Additional AWS infrastructure costs may apply. Use the AWS Pricing Calculator to estimate your infrastructure costs.
Dimension | Description | Cost/36 months |
|---|---|---|
Passage Retrieval calls | Call to the Passage Retrieval API | $0.01 |
Dimensions summary
This listing uses a single usage-based dimension: Passage Retrieval API calls, billed in units. You pay per call to the API, so cost scales with how many passage queries your AI applications send. There are no separate tiers, seat counts, or instance sizes to choose from. Your spend rises or falls directly with usage volume. The API returns relevant text passages from your indexed enterprise content to feed large language models and AI agents. Because pricing follows one metered unit, you can plan cost around expected query demand.
Top-of-mind questions for buyers
One unit is a single call to the Passage Retrieval API, also described as a passage query. Each request your application sends to retrieve text passages from your indexed content counts as one unit. Retrieval returns relevant text chunks with source links and relevance scores rather than full documents.
No. You are billed per API call, not per passage or document returned. A single call retrieves multiple relevant text chunks along with source links and relevance scores, and still counts as one unit. Your spend scales with the number of queries your applications send, not response size.
The pricing table meters each call to the Passage Retrieval API as one unit. Whether a call returns results depends on your query and index content. The available data does not specify handling of empty responses. Contact the vendor to confirm how such calls are counted.
coveo.com
Helpful?
Vendor refund policy
No Refunds
Custom pricing options
Request a private offer to receive a custom quote.
Legal
Vendor terms and conditions
Upon subscribing to this product, you must acknowledge and agree to the terms and conditions outlined in the vendor's End User License Agreement (EULA) .
Content disclaimer
Vendors are responsible for their product descriptions and other product content. AWS does not warrant that vendors' product descriptions or other product content are accurate, complete, reliable, current, or error-free.
Delivery details
API-Based Agents & Tools
API-Based Agents and Tools integrate through standard web protocols. Your applications can make API calls to access agent capabilities and receive responses.
Additional details
Usage instructions
API
To use the Coveo Passage Retrieval API (PR API) you will need the following items:
- Your Organization ID (e.g., mycompanyxyz123).
- An API key to authorize requests.
Follow these steps to configure Coveo in order to access your PR API endpoint and perform a test call.
1) Index your data
This step assumes that you're starting from an empty Coveo organization. Skip it if you already have content indexed in Coveo .
2) Configure the PR API
In the Coveo Platform Admin :
- Create a Passage Retrieval (CPR) model and associate it to a query pipeline.
- Ensure that the same pipeline is also associated with a Semantic Encoder (SE) model.
For complete step-by-step setup details, see: https://docs.coveo.com/en/o86c8334/build-a-search-ui/use-the-passage-retrieval-api
3) Create an API key
Create an API key and use it as a Bearer token in the Authorization header in the test call (step 4).
4) Make a test call (cURL)
In the following curl request, replace ORG_ID and API_KEY with your values.
curl -X POST <https://ORG_ID.org.coveo.com/rest/search/v3/passages/retrieve> -H "Authorization: Bearer API_KEY" -H "Content-Type: application/json" -d '{ "query": "", "additionalFields": ["clickableuri"], "maxPassages": 5, "localization": { "locale": "en-CA", "timezone": "America/Montreal" } }'
Expected response contains an items array with ranked passages and a responseId.
Please refer to Coveo's PR API documentation for more details on how to use Coveo's PR API.
For more details on the API itself, see the Passage Retrieval API reference .
Support
Vendor support
Ticket based https://connect.coveo.com/s/ 1.866.266.1583
AWS infrastructure support
AWS Support is a one-on-one, fast-response support channel that is staffed 24x7x365 with experienced and technical support engineers. The service helps customers of all sizes and technical abilities to successfully utilize the products and features provided by Amazon Web Services.
Similar products
Coveo unifies your enterprise content into a single permission-aware AI index, then delivers search, recommendations and grounded generative answers across commerce, service, website and workplace experiences, and to the agents you build on AWS.
Asana offers a Model Context Protocol (MCP) server, accessible via app integration, which allows AI assistants and other applications to access the Asana Work Graph from beyond the Asana platform. This server provides a way to interact with your Asana workspace through various AI platforms and tools that support MCP.
Structured agent access to Elasticsearch: Query indices using natural language via a standardized protocol -- no custom APIs needed
falcon-mcp enables seamless communication between AI agents and the CrowdStrike Falcon platform. Deployable directly onto Amazon Bedrock AgentCore, it provides programmatic access to Falcon data for agentic workflows and accelerating AI-native security automation.
Customer reviews
No customer reviews yet
Be the first to review this product . We've partnered with PeerSpot to gather customer feedback. You can share your experience by writing or recording a review, or scheduling a call with a PeerSpot analyst.