What Products and Capabilities Are Included in the Coveo AI-Relevance Platform?

Updated

Coveo's AI-Relevance Platform is organized into four product groups: search and discovery (AI Search, Conversational Search, Conversational Product Discovery), generative and retrieval capabilities (Generative Answering, Passage Retrieval API, Agentic AI & Retrieval), personalization and recommendations (AI Recommendations, Personalization), and supporting developer and indexing resources (Unified Indexing, Relevance Tuning, MCP Server, documentation). The platform is positioned for enterprise use across commerce, service, website, and workplace experiences, and Coveo publishes a pricing page and a free trial entry point for evaluation.

Search and Discovery Products

These are the core retrieval products that surface content and products from your indexed data.

Product What it does
AI Search Relevance-based search across enterprise content and product catalogs
Conversational Search Query-and-answer style search experience for enterprise content
Conversational Product Discovery Conversational discovery flow aimed at product finding in commerce contexts

The distinction matters when you're scoping a project: AI Search is the base retrieval layer, while the two conversational products layer a dialogue-style interaction on top of it. Coveo markets these as turning "every shopper's intent into conversions, and every customer question into clarity," which maps the first two to commerce and the third to service or support scenarios.

Generative and Retrieval Capabilities

This group covers how Coveo generates answers and exposes retrieval to other systems.

  • Generative Answering — produces generated responses grounded in your indexed content rather than only returning links.
  • Passage Retrieval API — returns relevant passages from your content, which is the building block for grounding generated answers or feeding another application.
  • Agentic AI & Retrieval — retrieval designed for AI agents, so an agent can pull from the same unified index rather than a separate pipeline.

If your goal is to add a Q&A layer to an existing site or support portal, Generative Answering plus Passage Retrieval API is the relevant pair. If you're building agents that need to retrieve from enterprise content, the agentic retrieval capability is the one to evaluate.

Personalization and Recommendations

  • AI Recommendations — suggests products or content based on behavior and relevance signals.
  • Personalization — adapts the experience to the individual user.

Coveo's page cites outcome metrics such as average order value, add-to-carts from listings, revenue per visit, and self-service success rate as the kinds of results these capabilities target. Treat those as the vendor's own framing rather than independently verified figures, and validate against your own baseline before committing.

Developer and Platform Resources

Beyond the user-facing products, the platform ships with the infrastructure needed to run and tune it:

  • Unified Indexing — consolidates content sources into one index so search, recommendations, and agents draw from the same data.
  • Relevance Tuning — controls for adjusting ranking and relevance.
  • Code Sandbox — a sandbox environment for testing.
  • MCP Server — a server for the Model Context Protocol, connecting the platform to MCP-compatible tooling.
  • Documentation and GitHub — developer references and code.

Coveo also lists integrations with Salesforce, SAP, Shopify, Adobe, Sitecore, Optimizely, ServiceNow, Zendesk, AWS, ChatGPT, and Agentforce. If your stack is on that list, integration effort is likely lower than a custom build.

How to Evaluate Before Buying

Coveo publishes a pricing page and a "Start a free trial" entry point, so you can begin with a trial and move to a sales conversation for commercial terms. A practical sequence:

  1. Start the free trial and connect one or two content sources to Unified Indexing.
  2. Test AI Search relevance against a set of real queries your users type.
  3. Add Generative Answering or Passage Retrieval API if you need answer generation, and check whether responses stay grounded in your content.
  4. Use Relevance Tuning to correct ranking problems you find.
  5. Check the integration list against your existing stack before scoping custom work.

The main decision point is which product group matches your problem: retrieval quality (search and discovery), answer generation (generative and retrieval), or experience adaptation (personalization and recommendations). Most enterprise deployments combine at least two.