Insights Studio
Platform Architecture & Capabilities
The full capability reference, plus the ClickHouse and PostgreSQL architecture, cluster-replicated rollups and read-only isolation behind the platform.
The full reference
Every capability, by area.
⇄
Getting the data in
Connect
- Connector framework — A pluggable backend with a 309-type source catalogue, covering databases, file sources and remote services.
- Ingestion — Extract-and-load jobs for bringing sources into the platform on a schedule.
- Transforms — A dbt-style transformation layer for modelling raw sources into something worth querying.
- Datasets — Multi-source datasets with materialization, and templates so a working pattern can be reused rather than rebuilt.
▦
Working with the data directly
Explore
- Database workspace — Query, page through results, and keep query history — a proper working surface rather than a dialogue box.
- Chart builder — Build a visual straight from a query result.
- Schema discovery — Inspect what is actually in a connection rather than relying on documentation.
- DB designer — Model and design schema within the workspace.
- CSV export — Get a result set out, because someone always needs it in a spreadsheet.
▤
Dashboards, scorecards and maps
Present
- Dashboard command center — The board-level view across dashboards.
- Scorecard editor — A full editor for scorecards, not a fixed template with slots.
- Visual gallery — 309 widget and chart types, including the Power-BI-class visuals people arrive expecting.
- Maps & routes — Real OpenStreetMap rendering with route display, as a native visual type.
- Scheduled reports — Automated report delivery as PDF over email, on a schedule.
✓
Making the numbers defensible
Trust
- Data quality — A self-serve wizard for the common tests, plus advanced tests for the cases the wizard cannot express.
- Data lineage — Connection-to-view lineage with a scope selector, built to stay readable rather than degenerate into a hairball.
- Rollup manager — User-defined pre-aggregations materialised into a columnar table with a relational cache in front.
- Per-view RBAC — Access control at the level of the individual view, with an audit log behind it.
- Alerting — Threshold alerts and anomaly detection over the modelled data.
- AI insights — Natural-language query and narrated anomaly explanation over the same governed datasets.
Under it
Platform & architecture.
| Analytical store | ClickHouse, with rollups replicated across the cluster |
|---|---|
| Relational store | PostgreSQL, including as the cache layer in front of materialised rollups |
| Connectors | A pluggable framework with a 309-type source catalogue |
| Modelling | dbt-style transforms, plus extract-and-load ingestion jobs |
| Materialization | User-defined rollup pre-aggregation, managed in-product |
| Visuals | A 309-item widget and chart gallery, plus OpenStreetMap-based geospatial rendering with routes |
| Governance | Per-view role-based access control, audit logging, data-quality tests and end-to-end lineage |
| Delivery | Interactive dashboards and scorecards, plus scheduled PDF reports over email |
| Read isolation | Dedicated read-only roles and grants across both stores |