Sundial Agents
The specialized agents you work with in Sundial, what each one does and how to pick the right one.
Sundial is agentic: instead of one do-everything chatbot, you work with a set of specialized agents, each grounded in your Context Engine so it reasons the way your team does. You pick an agent when you start a conversation, and every answer carries a trust signal so you know how far to lean on it.
Choosing an agent
Start a conversation with New chat in the sidebar, then select the agent that fits the job:
| Agent | Use it to | Availability |
|---|---|---|
| Analysis Agent | Ask questions about your business, from metric lookups to deep "why did this change?" analyses | Default, in every workspace |
| Data Quality Agent | Audit your metrics for freshness, breaks, and definition changes | Progressive rollout |
| Data Modelling Agent | Build and evolve your semantic layer and dbt models in natural language | Progressive rollout |
The Analysis Agent is selected by default and available to everyone. The Data Quality and Data Modelling agents roll out progressively; if they aren't in your New chat picker yet, ask your Sundial contact.
On the roadmap
A Storytelling Agent, which distills analysis into narratives and recommendations, is part of Sundial's direction for agentic analytics.
What the agents share
Every agent draws on the same foundation, which is what keeps answers consistent and trustworthy:
- The Context Engine: your semantic layer, playbooks, AI context, and warehouse metadata.
- Trust signals: a high/medium/low/unverified assessment on answers, based on whether the agent used governed definitions and known playbooks.
- The Library: where the Apps and Artifacts agents produce (reports, dashboards, data-quality reports) are collected, shared, and revisited.
Data teams tune all of this centrally, so improving context in one place makes every agent better.