The metric contract
Name, calculation, grain, time rule, source mapping, and permitted dimensions belong in the shared model.
Composable analyticsComposable analytics.
Composable analytics.
Change the view. Keep the definition.
Build an analysis for the question in front of you without redefining the metrics behind it. In Datus Studio, connect your data, bring reviewed business context into a project, and use the Agent to create, revise, and publish reports or dashboards. DOSI supplies executable semantic definitions where configured.
Northstar · Premier analysisIllustrative view
Shared metric
customer_marginReviewed definition · DOSIComposed HTML view
Premier customer overviewCustomer margin$13,821.893 Premier customers
The idea
One business definition. Many useful views.
Composable analytics separates the meaning of a metric from the page that displays it. A reviewed definition can serve a fixed dashboard, an exploratory HTML view, or a report. An agent can change the layout, filters, and chart choice for each question while the metric remains reviewable.
Reviewed metric
01Live dashboard
02Focused investigation
03HTML report
The analytical view
Cards, charts, tables, filters, and visual emphasis can change as the investigation moves.
See it in practice
Ask for a view. Then refine it.
Switch between two illustrative versions of the same Northstar Premier analysis. The presentation changes; the recorded customer-margin figures do not.
Definition held constant: customer_margin · reviewed DOSI metric
Request to Datus Agent
“Show Premier customer margin by customer, with total margin and delivered orders.”
Customer margin$13,821.893 customers · 21 delivered orders
| Customer | Margin | Delivered orders | Return rate |
|---|---|---|---|
| Maya Chen | $7,325.24 | 8 | 12.5% |
| Daniel Brooks | $3,308.16 | 7 | 0% |
| Sofia Rivera | $3,188.49 | 6 | 16.7% |
How Datus works
From a question to a view your team can use.
In a Datus Studio project, the Agent can start with an existing metric, supplied SQL, or both. It creates a report or dashboard, lets you refine individual sections, and gives your team a chance to review before publishing.
Existing metric or SQLDatus AgentPublished view
Choose the data contract
Start with a curated metric and its dimensions when one fits. An ad-hoc SQL query is another supported starting point, but it needs its own review.
Compose the page
Ask Datus Agent for headline cards, filters, charts, and tables. It saves an HTML dashboard and query templates for the configured source.
Review, revise, publish
Check the definition or SQL, filter meaning, time window, labels, and totals. Change one module as needed, then publish the reviewed artifact with the right access.
Choose the right output
Different audiences need different artifacts.
A generated page can be a working view or a maintained asset. Decide how it will get data and who must review it before you share it.
A view for a changing question
Build a focused HTML dashboard, follow a new breakdown, and keep revising the page while you investigate. See how this supports AI/BI analysis.
A live view for repeated checks
Publish a dashboard with filters and a configured data source. The Studio viewer queries through the Datus backend when an authorized reader changes a filter. Check data access and row policies before publishing. Explore agent-built dashboards.
A report with fixed results
Publish a reviewed HTML report as a stable, read-only viewer experience for a briefing or follow-up conversation. A report snapshot does not rerun its queries when opened. See how shared semantic definitions support reporting.
Fit your existing stack
Evaluate the workflow you already have.
Composable analytics is a stack decision as much as a dashboard decision. Start with your warehouse, the BI views people already trust, and the semantic approach your team can maintain.
Data sources
Check each source and semantic execution path separately; support varies by connector and feature.
Existing BI
Keep trusted views while you evaluate how shared definitions reach each tool. These examples are not a claim of direct integration.
Semantic approaches under review
Compare where metrics live, which tools can use them, and how changes are reviewed.
Questions to settle before choosing
Where do metric definitions live?
Can your team review the calculation, grain, joins, and source mapping once, then reuse them in the views that need them?
What stays in your current stack?
Verify the warehouse connector, the BI handoff, and whether a selected dashboard should remain where it is.
What runs when a viewer changes a filter?
Distinguish a live dashboard from a fixed report. Test the query path, access permissions, and row policies.
How do you change or move a view?
Check whether an Agent can revise a section without changing the metric, and review the resulting SQL or source references.
Start in Datus Studio
From connected data to a published view.
Use a Studio workspace to connect a data source, bring reviewed metrics into the project when available, and ask the Agent for a report or dashboard. Review the artifact, then publish it for the people who need it.
- 01 / Connect
Attach a supported data source to a project.
- 02 / Ground
Pull verified or certified metrics from Semantic Hub, or review the SQL used for an ad-hoc view.
- 03 / Compose
Ask the report or dashboard Agent to build and refine the view.
- 04 / Publish
Open the published viewer, test filters, and share with the right access.
Datus StudioIllustrative Studio workflow
Datus StudioNorthstar / Premier analysis
Reviewed metriccustomer_margin
Premier customer dashboardReady for review
customer_margin$13,821.89
Viewer · access controlledPublish dashboard
FAQ
Questions about composable analytics
What is composable analytics?
Composable analytics builds analytical views from reusable data access, metric definitions, query contracts, and presentation parts. A dashboard or report can change its layout without copying the underlying business calculation into every screen.
Does Datus need a DOSI metric to build an HTML dashboard?
No. Datus Agent can start from an existing metric, supplied SQL, or a combination. When a suitable reviewed semantic metric is configured, use it for a shared definition. An ad-hoc SQL chart still needs its own calculation and source review.
Can I edit just one chart or filter?
Yes. The documented dashboard agent can revise individual modules such as filters, KPI cards, charts, and tables. Review the resulting query and layout after each change.
Can I publish and share a generated dashboard?
In Datus Studio, a reviewed dashboard can be published to a viewer. Live filters run through the configured Datus backend for authorized readers; sharing is subject to the publication access and data policies. A locally generated HTML file does not bring that hosted runtime with it.
Which warehouse and BI tools can I keep?
Start with the data sources and BI views your team already uses. Datus Agent adapters, DOSI execution connectors, semantic-model exports, and BI handoffs have different support boundaries. Check the specific integration path and test permissions before planning a migration.
How should I compare Datus with Cube, MetricFlow, or warehouse-native semantic views?
Compare the model format, supported data sources, metric review process, query evidence, BI handoff, and who runs and shares the resulting dashboard. Datus combines Agent-generated views with reusable context; DOSI executes compatible Apache Ossie models where configured. Validate each option against your own stack.
Does this replace an existing BI tool?
It can work alongside one. Keep fixed dashboards for recurring monitoring and use generated views when the question changes. You can also move selected reports into a Datus workflow after checking definitions, permissions, and runtime needs.
Keep the meaning. Change the canvas.
Start with a business question and a definition your team can review. Let Datus Agent assemble the view that fits.
Build with Datus Agent