The Dawn of the Data Intelligence Engineer

AI is changing data work from processing a queue of requests to ensuring that people and agents can get reliable answers directly. Traditional analytics often involved a chain of handoffs: a business question exposed missing data, an engineer changed a pipeline, and an analyst built a report. The delay could make the answer less useful by the time it arrived. The author argues that newer models can now write SQL, inspect results, and revise their approach over multiple turns, making self-service analysis more practical than earlier AI tools or static dashboards. In one example, the author built an interactive presentation from four data sources using an AI agent, without writing code or SQL. Faster access can help employees ask follow-up questions while they still understand the problem, improving the speed and quality of decisions.

The proposed “Data Intelligence Engineer” focuses not mainly on writing queries or pipelines, but on making agent-driven analysis trustworthy. Agents can automate ingestion, transformations, and visualizations, while humans remain accountable for accuracy and business meaning. The role includes building clear data models, encoding definitions and relationships - such as what “ARR” or “revenue” means in a particular organization - and maintaining tests, or “evals,” that check whether agents answer known questions correctly. Broader access does create risks, especially when people act on incorrect data. The proposed safeguards are to let domain experts check results against their practical knowledge and to supervise agents in proportion to the potential consequences of their work.