Coveo launches RAG as a service for AWS agentic AI

Executive Summary

• Coveo Solutions Inc. has launched "RAG-as-a-Service," a new cloud-native offering designed to integrate with AWS agentic AI services.
• The service utilizes a Coveo-hosted Model Context Protocol (MCP) server to ground AWS tools (such as Amazon Bedrock AgentCore and Amazon Quick Suite) in enterprise knowledge, aiming to improve precision, security, and scalability for generative AI applications.
• The offering is being showcased at AWS re:Invent, with invitation-only early access available for developers to accelerate GenAI projects.

Key Details

• Product Offering: RAG-as-a-Service delivered via a fully managed, Coveo-hosted MCP server.
• Target Integration: Specifically designed to ground AWS agentic AI services, including Amazon Bedrock AgentCore, Amazon Bedrock agents, and Amazon Quick Suite.
• Core Capabilities: The service provides configurable tools including:
  • Passage retrieval: Returns relevant enterprise knowledge pieces to ground LLM prompts.
  • Answer: Generates precise answers from organizational data, powered by Amazon Nova.
  • Search: Retrieves ranked search results for context and exploration.
  • Fetch: Provides complete document text for complex reasoning tasks and deep research.
• Value Proposition: Aims to reduce retrieval complexity for developers, ensuring responses are grounded in factual, secure, and permission-aware data.
• Availability: Invitation-only early access is being offered to developers; the solution is being showcased at AWS re:Invent.

Notable Quotes

• "While LLMs have become widely available, their enterprise value depends on relevance, how effectively they can ground responses in factual, secure and permission-aware data," said Sebastien Paquet, vice-president of AI strategy, Coveo. "With our RAG-as-a-service offering, developers and enterprises can move faster, focus on innovation, and leave the retrieval complexity to us, the experts in enterprise relevance."
• "The new Coveo RAG-as-a-service offering helps connect AWS agentic AI services to enterprise-grade retrieval," said Eric Immermann, practice director for search and retrieval, Perficient. "By combining Coveo's proven relevance platform with models delivered via Amazon Bedrock, enterprises can deploy secure, grounded and high performing GenAI applications in record time."