One graph of who connects to whom, across every system.

Customers, accounts, devices and addresses arrive from several systems, duplicated and slightly different. Loaded into one store, they become a graph you can walk and rows you can report on, from the same records.

What it does.

Load once, with a receipt

Each CSV or JSON load records rows attempted, written and refused, and why. Each row's origin names its load, and a load with no quality check shows as unchecked, never as clean.

Shared attributes as a graph

Two customers with the same phone number or address are one hop apart. The exact-match index answers which records share a value without a scan.

The same records as rows

Report on the graph with SQL, or connect a BI tool over the Postgres wire protocol, without exporting a second copy.

Leading zeros kept

The CSV importer keeps values such as sort codes, phone numbers and postcodes as text.

On Enterprise, partly built.

Entity resolution

Rows matched across sources, scored against planted truth in our own tests.

Master data

A canonical model and a golden record that shows its working: each field records its source, the rule that chose it and the values that lost. Not yet: rebuilds write a new golden record instead of superseding the old one, and builds are not scheduled.