The Analytics Engineering Podcast Agentic coding in analytics engineering (w/ Mikkel Dengsøe)
37 snips
Sep 7, 2025 Mikkel Dengsøe, co-founder of SYNQ, dives into the world of agentic coding and its transformative impact on analytics engineering. He shares a hands-on project using tools like Cursor and Snowflake, discussing where agents excel—like in staging and lineage checks—and where they pose risks, such as in BI chat for novices. Mikkel also emphasizes the shift from traditional dashboards to actionable insights, underlining the need for human expertise in AI integrations and the proactive evolution of data observability.
AI Snips
Chapters
Transcript
Episode notes
From Zero-To-One To Observability
- Mikkel started his career building a zero-to-one data warehouse at a Danish shipping company and later worked at Google on curated ad data.
- He then scaled analytics at Monzo and co-founded Sync, a data observability platform.
Low-Risk Entry Points For Agents
- Mikkel argues most data teams should experiment with AI agents because risk is low for internal modeling and testing workflows.
- He says agents boost productivity while keeping humans in control for these tasks.
Modeling Real Estate Data With Cursor
- Mikkel loaded raw JSON and XML into Snowflake and modeled through staging, mods, and metrics using Cursor and the dbt MCP plugin.
- He fed sample CSV extracts into Cursor so it could inspect raw data while writing models.