CallFlowGraph: Journey-Aware Graph Agents for Support Analytics

Session Track: Data Intelligence

Session Time: 06:00 am - 06:30 am November 6

Session description

“I was asked to explain why thousands of customers were still calling—even after using the app or chatbot. But our tools only showed individual tickets, not the story behind them. We needed more than dashboards. We needed journeys.” In this session, I’ll walk through CallFlowGraph—a Neo4j-powered customer journey graph enriched with LangGraph-based AI agents that help operations teams understand how and why users escalate across support channels. We modeled accounts, intents, channels, and timestamps as an explainable interaction graph, enabling queries such as: Which intents drop off most between digital and call? What does a “frustrated journey” look like over 24 hours? Who tried chat and failed, then called within 30 minutes? With this setup, support managers ask natural language questions, and behind the scenes, a planner agent dynamically invokes Cypher queries and auto-generates visual plots. The outcome? Journey-level answers—not just ticket counts—with traceable, context-aware logic. If you’ve struggled with fragmented support data, failed digital containment, or "guess-based" escalation patterns, this talk will show how combining graph structure with autonomous agents enables a new way to ask, answer, and act on complex CX questions—without building custom apps or dashboards. Whether you’re in insurance, telco, or any high-touch service domain, CallFlowGraph makes your customer interactions not just observable—but narratable.