Building a Customer Intelligence Brain: How GraphRAG Turns Qualitative Data into Deliverables
Session track: AI Engineering
Session time:
Session description:
When an AI agent can traverse a structured customer intelligence graph through purpose-built MCP tools, the graph can become an evidence layer for generating decisions and business artifacts - not just answers.
The same evidence base that produces a pain prioritization heatmap can also generate a go-to-market strategy, a product brief grounded in real customer jobs-to-be-done, a personalized onboarding plan for a specific customer segment, or a value proposition that traces every claim back to the underlying interview evidence. Not as summaries - as structured, defensible, traceable outputs.
In this session I will show how a single customer intelligence graph generates categorically different outputs using an AI Agent connected to Neo4j via MCP: a strategic opportunity brief, a needs-based product concept, a segmented value proposition, an individualized customer journey plan, and a board-ready recommendation document.
Each one traverses the same graph differently. Each one is generated in under two minutes. Each one traces every claim to source evidence.
The session will cover what makes graph-native generation categorically different from prompting an LLM with raw transcripts, why typed semantic relationships and explicit evidence paths provide richer context than keyword similarity alone, and how the compounding nature of a knowledge graph means every new interview makes every future output sharper.
Attendees will see:
• how interviews, source excerpts, customer jobs, pains, outcomes, and evidence relationships are represented in Neo4j;
• how purpose-built MCP tools expose the graph to an AI agent;
• how generated findings are linked back to source excerpts and graph paths.
If you attended or watched my April 2026 session, this is the technical next chapter to Building a Customer Intelligence Brain: How GraphRAG Turns Qualitative Data into Decisions, presented at NODES AI 2026. If you are new - everything you need will be covered from first principles.
Speaker
Eckhart Boehme
Managing Director, unipro solutions GmbH & Co. KG
Eckhart Boehme is the founder and managing director of boutique consulting firm unipro solutions and an internationally recognized Jobs to Be Done expert. Eckhart was the initiator and co-developer of The Wheel of Progress® Canvas - a tool for structuring qualitative customer research. He is the developer of the Customer Progress Design method. Eckhart is a frequent guest at business podcasts and has taught classes to Master Students at the Wiesbaden Business School and Michigan State University. Together with his business partner he developed a method for building customer-intelligence knowledge graphs.
