idbigintInternalcustomer_idbigintInternalregion_idintPublictotalnumericConfidentialcreated_attimestampInternal
Your data,
understood.
Tables, files and databases become one layer you can query, model and act on. Ask in SQL or in plain words, give your data meaning with an ontology, and keep the audit trail where you can see it.
Tables that live with your app.
Every app has its own Data Studio: fast columnar tables that flows write to, with search, SQL and forecasts built in.
Lance tables
Columnar storage built for analytics and AI. Flows create and fill tables as they run.
Vector & hybrid search
Find rows by meaning, by keyword or by both, with 16 index types from B-trees to HNSW.
A SQL workbench
Query tables, ontologies and shared contracts in one place. Results as a table, chart, graph or JSON; save queries as views.
Explore & edit
Browse objects, follow relationships and edit properties, with conflict detection and permissions.
Forecasts
Predict any time series with five foundation models, evaluated against a held-out window.
Flux knows your data
Ask Flux to build tables, write SQL or Cypher and chart the answer. It shows the query it ran.
One question.
Every source.
Register your databases, lakes and files in one session and join them with SQL. No copies, no pipeline to maintain first.
idint32Publicnameutf8Public
-- Revenue per region this quarterSELECT r.name AS region, SUM(o.total) AS revenueFROM shop.orders oJOIN app.regions r ON r.id = o.region_idWHERE o.created_at >= DATE '2026-10-01'GROUP BY r.nameORDER BY revenue DESC;| region | revenue |
|---|---|
| Americas | β¬538,410 |
| EMEA | β¬412,300 |
| APAC | β¬201,950 |
- PostgreSQL
- MySQL
- SQLite
- DuckDB
- ClickHouse
- Oracle
- BigQuery
- Athena
- FlightSQL
- Delta Lake
- Iceberg
- Parquet
- CSV
- JSON
- Lance
Give your data a model of your business.
An ontology names the things your business runs on and how they connect. It's metadata on top of your tables, so nothing is copied, and you can keep several over the same data.
- Relationships are proposed from your column names; no foreign keys needed.
- Explore the graph, find paths and run Cypher queries.
- Actions turn changes into governed, versioned operations with validated inputs.
- Share an ontology with other apps as a contract, without sharing the data behind it.
From what happened
to what's next.
Forecast, classify, cluster and detect anomalies without leaving the app, and without a Python stack to babysit.
- Five forecasting models, from TimesFM 2.5 to Chronos-2, running natively on the CPU.
- Training nodes for classification, regression, clustering and dimensionality reduction.
- Auto training for tabular, forecast, vision and anomaly tasks, as resumable background experiments.
- ONNX inference for images, OCR, faces, poses, audio and named entities.
Trust it in an audit.
Data you can't explain is a liability. These are built in, not bolted on.
Change a column's label and watch the table's label follow. These are the levels and wording the app uses.
idint64nameutf8emailutf8countryutf8segmentutf8credit_limitfloat64
- 01
Sensitivity labels
Label every column public, internal, confidential or restricted, and flag personal data. Tables show the strictest label.
- 02
Governed actions
Changes run through pinned flow versions with validated inputs, checked again at the moment they run.
- 03
Write-once storage
Keep records on immutable object storage with retention, legal hold and a hash for every version.
- 04
A tamper-evident trail
Signed, chained records of every change, exportable to your SIEM and verifiable offline.
Dashboards are the last mile.
Once the data layer is right, dashboards are easy: bar, line, heatmap, Sankey and more, bound to live queries in any page. Keep governed KPIs in SQL, and let AI explain them.
See Flow-Like for BIPut your data to work.
Start with a table, a CSV or your production database. Data Studio is in every app.