Describe the pipeline. Deploy it live. Move the data under policy.
Catalyst is a transport-agnostic, declarative data-movement and transformation platform for moving disparate data across network and domain boundaries. Describe your pipeline in YAML, deploy it live to any node in your fleet, and Catalyst moves the data durably, observably, and under policy.
Transcript
Transcript to follow.
Moving data is where programs stall.
Data sits in systems that were never designed to talk to each other, across networks and domains that were never designed to be crossed. Every new connection becomes its own project, with its own bespoke code, its own failure modes, and no shared account of what moved, when, or under what rules.
Declarative movement, deployed live.
You describe what should move and how it should be transformed. Catalyst runs it: ingest from a source, transform through a pipeline, store durably, push to multiple sinks. Changes deploy to running nodes without an image rebuild or a restart, and the node reconciles automatically when it boots.
How Catalyst Works
A pipeline ingests from a source, transforms through its stages, stores durably, and pushes to the sinks that need it. Click any workspace to connect or disconnect.
What defines Catalyst
The runtime unit is a single Go binary with no external dependencies, built for the contested edge rather than retrofitted to it.
One binary, no dependencies
The runtime unit is a single Go binary with no external dependencies. It ingests from a source, transforms through a pipeline, stores durably, and pushes to multiple sinks. Nothing else has to be present for it to run.
No inbound control port
Nodes dial outward to the fleet manager and hold the session. No node exposes an inbound control port, so the node does not need to be reachable to be managed.
Air-gap-native
Registry commands can populate a node's artifact store directly on disk with no fleet manager present, and a single flow can run offline. Deployment state survives restarts. Air-gap operation is the design point, not a retrofit.
Live deployment
Programs deploy live to running nodes with no image rebuild and no restart, and reconcile automatically on node boot. You change what moves without taking the pipeline down.
Degrades gracefully
Every inter-stage handoff is a durable bounded queue with selectable strategies and retention limits, so the system degrades predictably under load instead of dropping data or buffering without bound.
Technical foundations
The pieces that make declarative movement dependable in environments that punish assumptions.
Declarative pipelines
Pipelines are described in YAML rather than written as bespoke integration code. The description is the deployable artifact.
Five stage roles
A pipeline is composed from five stage roles, covering ingest, transform, storage, policy, and delivery. Roughly three dozen first-party plugins ship across all five.
Durable bounded queues
Each handoff between stages is a durable bounded queue with selectable strategies and retention limits. Backpressure is a design decision you make, not an accident you discover.
Content policy operators
Policy operators redact sensitive patterns and filter fields. They terminate the raw payload, so unredacted bytes cannot pass the policy stage.
Out-of-process plugins
Any stage can be an out-of-process plugin over a versioned contract, so custom or NDA'd integrations stay out of the core binary and a plugin crash cannot take down the node.
Observable by default
Movement is observable as it happens: what ran, what moved, and what the policy stage did to it.
Deployment Model
Catalyst runs where the data already is, including places with no connectivity to anything.
Fleet-managed
Nodes dial outward to the fleet manager and hold the session, so programs can be deployed live across the fleet without any node exposing an inbound control port.
On-premises
The runtime is a single Go binary with no external dependencies, so a node runs on customer-managed hardware without pulling in a supporting stack.
Air-gapped
Registry commands populate a node's artifact store directly on disk with no fleet manager present. A single flow can run entirely offline, and deployment state survives restarts.
Contested edge
Built for the contested edge rather than retrofitted to it. Bounded queues and durable handoffs mean intermittent connectivity degrades throughput rather than losing data.
Services That Deliver Catalyst
Products don't deliver outcomes. People do. Orbis engineers build the integrations, deploy the nodes, and design the pipelines that make Catalyst operational for your mission, not just installed.
Source integration
Connecting your existing sensors, systems, and data sources into Catalyst pipelines, including custom plugins that stay out of the core binary.
Delivery integration
Getting data into your mission applications, dashboards, and decision systems in the format and cadence your operators need.
Node deployment
End-to-end deployment into your environment, from pipeline design through operational handoff, including air-gapped and disconnected configurations.
Architecture and pipeline design
Designing what should move, how it should be transformed, and what the policy stage must enforce before anything leaves a boundary.
The organization that governs how its data moves, governs how fast it can act.
Describe the pipeline in YAML. Deploy it live to any node in your fleet. Move the data durably, observably, and under policy.