Data Modelling Agent
Build and evolve your semantic layer and dbt models in natural language, every change ships as a pull request.
The Data Modelling Agent builds and evolves your data model the way you'd ask a data engineer to. You describe what you want in plain language, "add a weekly-active-users measure", "create a customer_region dimension", "set up a daily pre-aggregation for orders", and the agent writes or edits the underlying dbt models, semantic models, and metrics.
It works in a Git-backed workspace, grounded in your actual repository and warehouse rather than guessing from table names, so its changes fit how your data is really shaped.
Progressive rollout
The Data Modelling Agent is being rolled out gradually. If it isn't in your New chat picker yet, ask your Sundial contact.
Where to find it
Start a conversation by clicking New chat in the sidebar and selecting Data Modelling Agent.
Every change is a pull request
The agent never edits your model silently. Its deliverable for any change is a pull request:
- You describe the change in natural language.
- The agent inspects the current definitions and drafts the edits, semantic YAML plus any companion dbt model, and validates them.
- It opens a pull request with the changes for you to review.
- You review and merge. Git sync picks up the commit and the change takes effect.
Because the model lives in Git and every change is a PR, your semantic layer gets normal code review, history, and rollback.
What it can do
- Model and measure: create and edit semantic models, dimensions, measures, and pre-aggregations.
- Explore sources: inspect source tables and sample data to ground each change.
- Trace lineage: follow upstream/downstream dependencies to understand the impact of a change.
- Run backfills: rebuild models over historical ranges after a change.
- Debug pipelines: diagnose and re-run failed pipeline jobs (where connected).
It's scoped to your data model and its pipelines, not BI dashboards or business strategy. When a request falls outside that, it tells you what it can do instead.
Getting started
New to the semantic layer? See Initial Setup for how to seed your models from an existing dbt semantic layer or author them from scratch, then evolve them with this agent. The full modelling concepts live in the Semantic Reference.