DAX queries by agents

[JuanSombrero23]

Silly(?) question from my side. There's mutliple ways to have agents launch DAX queries against an online semantic model. fab CLI, PBI MCP, SemPy, and now te CLI.

My gut feeling tells me now that te CLI becomes the preferred way because

• i could leverage BPA checks against the queries
• errors would be captured by the CLI before being sent

The quality of the DAX by itself is dependent on the agent and the skills, not related to the tool you use.

Does this make sense? Anything I miss?

Thx

[mlonsk]

Hi @JuanSombrero23
Glad to hear you find the TE CLI useful for agentic use cases.
Overall, the quality of the DAX an agent generates depends on several factors, including the model, context, prompt, tools, and the agent's environment.
The TE CLI can improve DAX quality by, for example, parsing it directly with the TE CLI's semantic analyzer, thereby enforcing quality control over the generated DAX queries.