Explicit Measures Podcast 466: Data Agents & Semantic Models
Oct 9, 2025
Dive into the world of data agents and discover if semantic models are the new MVP of data sourcing. Explore the integration of Microsoft Dynamics and Power BI, revealing their strengths and reporting capabilities. Learn how data agents generate queries through natural language processing and why they differ from Copilot. Engage with the debate on semantic models for enhancing large language models and gain insights into crafting effective prompts. Finally, get practical tips on testing agents while managing capacity.
AI Snips
Chapters
Transcript
Episode notes
Build Agents With Clear Instructions
- Use data agents to provide conversational Q&A over your data without writing code.
- Configure agents by writing clear instruction prompts and context for reliable responses.
NL2Query Engines Vary By Source
- Data agents translate natural language into SQL, DAX, or KQL depending on the data source.
- SQL translations are strongest today while DAX and KQL support is improving.
Prefer Task-Specific Agents
- Treat the agent like a configurable tool, not a one-size-fits-all copilot.
- Create task-specific agents with tailored instructions when you need consistent behavior.