Talk to your data: AI-powered conversational analytics with the dbt MCP server скачать в хорошем качестве

Talk to your data: AI-powered conversational analytics with the dbt MCP server

Ask three different chatbots for “revenue this quarter,” and you may get three different numbers. Trust erodes, adoption stalls, and your data team inherits the mess. Without shared definitions and lineage, chat interfaces can create costly rework and compliance risk. If you can’t trace the metric and see how it was produced, you probably shouldn’t trust it. That’s the challenge Norlys, Denmark’s largest energy-telecom group, set out to solve with LEAP Data & AI Consulting. Together, they designed and launched a production-ready conversational analytics tool powered by the dbt Model Context Protocol (MCP) server and the dbt Semantic Layer. By exposing definitions, tests, and lineage to their AI system, Norlys delivers fast, verifiable self-service insights to every business user—without creating new risks for the data team. Whether you’re exploring AI use cases or preparing to ship conversational analytics into production, this session offers a practical playbook for turning today’s hype into real business value. What you’ll gain from watching: -Building reliable AI chat experiences: Practical lessons from Norlys and LEAP on designing a conversational analytics interface that works in production -See it in action: A live demo of Norlys’ conversational analytics experience—how plain-English questions become accurate, trusted answers -Operational takeaways: Guidance on scalability, reliability, and integration considerations for AI-powered BI -Team impact: How conversational analytics reduces analyst workload and expands business access to data -Next steps: Where to start if you want to experiment with the dbt MCP Server and the dbt Semantic Layer in your own stack Meet your speakers: -Søren Meincke Persson, Head of Data Engineering at Norlys -Jonas Munk, AI & Data Strategy Lead at Leap, Data & AI Consulting -Thomas Grabowski, Product Manager, AI at dbt Labs Chapters: 0:00 Welcome and Introduction 01:45 Event Overview and Housekeeping 04:00 Understanding Structured Context in AI 07:19 Introducing the Co-Presenters 09:53 Company Background and Vision 13:43 Challenges and Strategic Decisions 17:48 Building a Unified Data Platform 29:00 Metric-First Approach 32:58 Metric-First Approach and Data Lineage 34:43 Enforcing Common Data Definitions with dbt Semantic Layer 35:06 Aligning Metrics with Business Goals 36:03 Chat Interface for Data Interaction 37:27 Scaling AI and LLM Integration 38:23 POC and Future Possibilities 44:44 Modular Approach and Centralized Interface 46:15 Final Thoughts and Recommendations 50:24 Q&A Session