Merchandisers act on performance questions without leaving their workspace, and nothing the agent proposes applies without approval.
Conversational Search and Product Discovery
Customers, employees, and buyers reach grounded answers and product decisions by describing what they need.
A conversational pair that answers questions with cited sources and assembles product pages from shopper intent.
What the agents do
Chooses its own retrieval steps and writes cited answers, and on shopping pages classifies intent and assembles its own layouts.
- Built by
- Coveo, which builds and runs both agents on its own search and discovery platform
- Contributed by
- Coveo (Conversational answers and product discovery on its own platform)
- Industry
- Cross-industry, Technology, Grocery, Automotive, Healthcare
- Audience
- B2B, B2C
01
The problem
Why this exists
Customers, employees, and buyers usually know what they need to accomplish. What they cannot do is reach the answer or the product without work: the information is split across systems, so they piece it together, run the same query again in different words, and compare options by hand.
Read more
Asking for help in plain language only helps if the reply is grounded in what the organization already knows and respects the access rules it already sets. Otherwise the answer is confident, plausible, and unusable in a regulated workplace.
Shopping intent is harder still, because a shopper rarely states it. Reading that intent and assembling products, refinements, and product information into a page a person can explore takes as much manual configuration as it saves.
02
Outcomes
What changes when this is in place
- Forcepoint reported a 14% improvement in self-service success after three months, measured by sessions that did not go on to open a case.
- It reported a 60% improvement in explicit case deflection after five months, on case-form visits that ended without a submission.
- Both figures come from Coveo Relevance Generative Answering, an earlier generation of the answering technology this entry describes.
03
How it works
Architecture
Both capabilities run as separate agents on the vendor's platform, each reached through its own endpoints, with a shared orchestration layer on the roadmap rather than in the product. The answering side interprets the question, chooses among retrieval tools, reasons across knowledge that has already been indexed, and writes a reply with citations, holding context from one turn to the next.
Read more
The discovery side classifies shopper intent, applies a policy graph, retrieves catalog data, and selects a layout of products, refinements, and product information for the retailer's own components to render. Product-linked content can supplement the catalog.
Retrieval respects the permissions and query rules an organization already maintains, so an answer never reaches past what the person asking could open. Brand context shapes tone and behavior, and layouts are built from components the customer designed rather than vendor templates.
Tech stack
Coveo Search AgentCoveo Discovery AgentCoveo retrieval modelsCoveo Commerce APIAmazon Bedrock
Next step
Request an intro
Coveo is behind this solution. Ask about putting cited answers and product discovery into one conversation, and the MACH Alliance will put you in touch with the person who can answer.
Not ready to talk to anyone? MACH Insider is the Alliance mailing list. New solutions, program openings, and the research behind them, once a month.
Everyone credited on this solution
- CoveoConversational answers and product discovery on its own platform
Every organization named here has a page listing the rest of its work on the Hub.
Related solutions
Make products easier to find, compare, and buy by completing the catalog details that search engines and AI assistants need.
Agentic Product Data Enrichment
Support answers arrive inside the product, so integration teams stop leaving their work to search three sources or wait on a ticket.
Patchworks Answers Agent
Verification
Contributor submission
The detail on this page comes from the contributor's own account of their own work. The MACH Alliance reviews a submission before publishing it and does not independently verify the figures in it.
What this page does not establish
- No measured outcome for the conversational combination itself is published: the Forcepoint figures predate its conversational capabilities.
- Product discovery is in Early Access with a September 2026 general availability target, and the submission asks for that rollout status to be confirmed.
- The submission names no customer running both capabilities together, so no deployment size, storefront, or team is established here.
Where this came from
Solution detail is drawn from Coveo's submission to the hub, a Conversational Search and Product Discovery listing. The interim results are attributed in the submission to Forcepoint's deployment of Coveo Relevance Generative Answering.
Status: Staged rolloutAdded Oct 30, 2026A listing is an editorial decision, not a MACH Alliance certification.
Built something worth showing?
Put your agentic solution in front of the enterprise leaders who use the hub to decide what to build next. Submissions are open to MACH Alliance Members, Collaborators, or Ambassadors.