Nexma

Nexma SyncEngine

Connect the world's data feeds

Nexma SyncEngine is the ingestion layer of the platform. It discovers, connects, normalizes, and continuously syncs hundreds of live spatial, sensor, and enterprise sources into the world model through one synchronization layer — so every decision runs on current, unified data instead of brittle one-off pipelines.

Every source speaks its own format, cadence, and projection. The conventional answer is a custom connector per source that breaks the moment an upstream endpoint changes, landing data in silos that never reconcile into a single picture. SyncEngine replaces that with a catalog of connectors and a normalization path into the Nexma DataStore.

Core concepts

  • Source. Any external system SyncEngine can read — a live spatial feed, a sensor stream, or an enterprise system of record. Each source has a connector, a cadence, and a credential.
  • Connector. The adapter that translates a source's schema, units, and projection into the world model on ingest. Hundreds are prebuilt; a connector framework covers proprietary or niche sources.
  • Normalization. The step that maps each incoming record's format and projection to a typed entity, then reconciles identifiers so the same real-world object never appears twice.
  • Sync. The continuous process that keeps a source live — polled or streamed on its natural cadence, applying only the changes that arrived since the last run.
SyncEngine connects the breadth of sources into the world model. For high-frequency, multi-protocol telemetry — fleet positions, live sensor readings — pair it with the Nexma Event Broker. They work together.

How it works

A source moves through four stages, from a catalog entry to a typed, current layer in the DataStore.

StageWhat happensResult
DiscoverBrowse the source catalog and enable a connectorSource registered
ConnectAuthenticate with managed credentials and tokensLive connection
NormalizeMap format, units, and projection; reconcile identityTyped records
SyncPoll or stream on the source's cadence; apply changesAlways-current world model

Once a source is syncing, every surface reads the same reconciled data — the Globe, tables, and Jax all see the present, not last night's batch.

Connectors

SyncEngine ships connectors across the categories teams integrate most. The catalog grows continuously, and custom connectors cover anything not yet prebuilt.

CategoryExample sourcesTypical cadence
AviationLive flights, airport statusSeconds
MaritimeVessel positions, port congestionSeconds to minutes
WeatherRadar, severe-weather alerts, forecast surfacesMinutes to hourly
Earth observationWildfire detection, vegetation indices, imageryMinutes to daily
MobilityTraffic, transit, freight flowsSeconds to minutes
CadastralParcels, zoning, permitsDaily to on demand
EnterpriseCRM, ERP, asset and work-order systemsStreaming or scheduled
MarketsCommodity prices for relevant verticalsMinutes to daily
CustomProprietary APIs, on-prem sensor streamsSource-defined
Sources are enabled per project. Ones you do not enable never touch your project.

Connect and normalize

A connector does more than pull bytes. On the way in, it translates each source's schema, units, and projection into the world model so records land typed and consistent. When a project has an active Ontology, incoming records are typed against it — usable the moment they arrive, queryable by entity and property rather than by raw column.

Identity reconciliation matches identifiers across sources, so a vessel reported by two feeds resolves to one object rather than two duplicates. Credentials and tokens are managed centrally, keeping secrets out of scattered scripts.

Sync, scheduling, and health

Each source runs on its natural cadence — a scheduled poll or a live stream — with no manual jobs to babysit. Sync is incremental: only the records that changed since the last run are applied, which keeps the layer efficient even across very large datasets and avoids re-ingesting unchanged history.

Every source carries health telemetry:

  • Freshness — how recently the source last delivered data.
  • Throughput — the rate of records flowing in.
  • Failures — connector errors, auth expiry, or upstream outages, with alerts when a feed goes quiet.

This makes a stale or broken feed visible immediately, rather than discovered when a decision turns out to have run on missing data.

Where it fits

SyncEngine is one of two ingestion paths into the platform. It handles discovery and breadth; the Event Broker handles real-time telemetry volume. Both write into the DataStore, where the Ontology gives every record meaning and Jax reasons over it through the eight generic primitives.

Where to go next