Working with Dawn

Choose the right place to explore, build, and operationalise analytics work.

Dawn is an AI-first analytics workspace. You can begin with a question, then turn the useful parts of the work into assets your team can inspect, reuse, and run again.

The core loop

Most work in Dawn follows a simple loop:

  1. Choose the data and context. Start with a dataset and the business question you want to answer.
  2. Explore with the agent. Ask for analysis, challenge assumptions, and narrow the work as you learn.
  3. Save the useful result. Turn repeatable transformations into a pipeline, an audience-facing view into a dashboard or report, and durable knowledge into Context.
  4. Operationalise it when needed. Create a job or schedule when the same work should happen again.

Where work happens

Conversations

Conversations are the fastest place to investigate a question. Add the relevant dataset or existing asset to the conversation, state the decision you are trying to make, and ask Dawn to show its reasoning and next steps.

Workspace assets

Use the workspace to organise the things your team builds:

  • Datasets hold source data for analysis.
  • Pipelines capture repeatable preparation and transformation work.
  • Dashboards support interactive monitoring and exploration.
  • Reports present a narrative, a conclusion, or a regular update.
  • Models hold data-science and machine-learning work where applicable.

Agent work

Use Jobs for a defined task that should run on demand or on a schedule. Use Context to give agents durable information about your organisation, data, definitions, and working conventions.

A good way to start

Keep the first request focused. Name the data, the period or segment you care about, and the decision the result should inform.

Using the sales dataset, compare this quarter with the same quarter last year.
Explain the largest changes by region and product, flag unusual movements, and
recommend the two questions the commercial team should investigate next.

Once the answer is useful, decide whether it should remain an exploration, become a shared dashboard or report, or run again as a job.

Next steps