Data model

From tables to a living model of your operation.

Every picture in the twin stands on a model: the vessels, berths, cargo, people and port calls your operation actually runs on, linked together, traced back to source and kept through time. Follow MV Kestrel into Harbour Watch and watch the model build, explain itself, rewind and branch.

01 · Ingest & map

Three systems, three rows — one Kestrel.

Information about the same ship sits in three different systems. Tessara recognises it and joins it into one record.

The AIS feed knows an MMSI. The berth schedule knows a code. The customs manifest knows a container and a port call. None of them knows Kestrel. Tessara maps each source's columns onto object types once, and from then on every row that lands resolves into the same governed objects.

Declarative mappings from columns to object types, properties and links — versioned like code
Identity resolution across sources, so one vessel is one object however many systems describe it
Connector agents inside each source network; nothing leaves your perimeter
02 · The ontology

Objects, links and properties — the shape of the operation.

Your operation as the things you actually manage — and how they connect.

The ontology is the vocabulary your operators already use: a vessel is assigned to a berth, carries cargo, makes a port call; cargo is consigned to a person. Select any object to see its properties and what it links to.

Typed objects with typed links — traverse from Kestrel to Okafor in two hops
Property-level access policy and classification markings
The same model drives search, the analyst, the twin and every dashboard
03 · Provenance

Ask any value where it came from.

Select any value to see exactly where it came from.

Kestrel's ETA says 14:20. Why? Every property carries its lineage: the source row it was read from, the sync job that delivered it and the pipeline version that shaped it. Select a property to trace it.

Lineage on every property, not just every dataset
Pipeline versions pinned, so an answer from last week can be reproduced exactly
Auditor-defensible from decision back to source row
04 · Time

The model remembers — rewind to any moment.

See what the picture looked like at any moment in the last day.

Objects don't just have a state, they have a history. Drag through the last 24 hours to see Kestrel go from at sea to inbound, alongside, discharging and finally under customs hold, with her berth assignment and ETA as the model knew them at each moment.

Bitemporal history: what was true, and when the model learned it
Replay an incident exactly as operators saw it at the time
History is governed by the same access policy as the present
05 · What-if

Branch the model. See the knock-on effects before they happen.

Try a change in a safe copy of the model and see what it would affect.

Fork the live model into a scenario and change the world: take Berth 12 out of service, or let the tide run 40 minutes late. Tessara pushes the change along the links, from berth to vessel to cargo handover to crew shift, and shows you the differences side by side with live.

Scenarios are branches of the model; live data is never touched
Consequences follow the links, so nobody has to remember the dependencies
Compare branches, then hand the chosen plan to operators

Model your operation

Harbour Watch is our scenario; the model is built from your systems. If one governed, traceable model of your operation would change how you decide, talk to us.

Get a demo