System architecture

Dawn is organised around the work your team creates: data, the assets built from it, the agent that helps with analysis, and the controls that make the work shareable and governable.

The main layers

Data

Datasets provide the inputs for analysis. They can begin as a file upload or be supplied through a connection or integration. Pipelines capture repeatable preparation and transformation work so the same logic can be run again as data changes.

Analytics assets

Dashboards, reports, models, and other workspace assets turn data into something people can use. These assets provide a durable home for work that should be reviewed, shared, or improved over time.

Agent work

Dawn uses the relevant data and workspace context to help users investigate questions and create assets. Jobs package a defined piece of agent work so it can be run on demand or on a schedule. Context holds reusable business knowledge, definitions, and conventions.

Governance

Permissions control who can access an organisation and its workspaces. Teams should also establish clear ownership for important datasets, metrics, pipelines, and recurring jobs.

A typical flow

1. Bring source data into a dataset.
2. Prepare it in a pipeline when the transformation needs to be repeatable.
3. Explore the data with Dawn and validate the result.
4. Publish a dashboard or report for the intended audience.
5. Schedule a job when the analysis or update must recur.

This structure lets exploratory work mature into an operational workflow without losing the original business context.

Design principles

• Keep raw inputs and reusable transformations distinct.
• Give important assets a clear owner and purpose.
• Record metric definitions and business rules in shared Context.
• Validate results before they inform decisions or automated actions.

For access guidance, see Permissions & security.