Platform
Four products. One platform underneath.
VMS + Omniscient, School OS, Ball Track, and OrgBrain are not four separate builds. They are four products on one engineered software platform: the same inference layer, the same access control, the same audit record, and the same three deployment surfaces. Your data never leaves the boundary you set.
The deployment surface
Three places the software runs
Every product ships to all three. The surface changes where the software runs and who holds the hardware. It does not change the release, and it does not change what the product does.
On-premise
Runs on hardware inside your building.
Inference, storage, and logs all sit on servers the customer owns, behind the customer's own network boundary. Nothing is sent outward to be processed. This is the surface that makes the platform usable in sites where recording and record-keeping are already regulated.
Private tenant
Runs inside your own cloud account.
The same software deploys into the customer's own cloud subscription, Azure-first, where the requirement is that data stays in an account they control rather than on hardware they control. Identity, keys, and logs remain the customer's.
Edge
Runs on a node next to the sensor.
A single self-contained GPU node placed beside the cameras or the video feed, for sites with thin uplink or no uplink at all. It processes locally and sends only events and results onward, never raw footage.
The shared layers
What every product is built on
GPU inference
One accelerated inference layer serves every detection and generation workload on the platform. Ball Track's frame-by-frame tracking, Omniscient CV's real-time detection, and OrgBrain's model serving are the same pipeline pointed at different models.
Private model serving
Models run on hardware the customer controls, with weights held locally. No prompt, frame, or document is sent to a third-party model provider, which is the property the whole platform is designed around rather than a setting bolted on at the end.
Audit and access control
Every product writes the same shape of record: who asked, what the software returned, which source or camera or student record it came from, and when. Access is resolved against the roles the organisation already enforces, not a second set of permissions living beside them.
Integrations
Standard connectors into the systems a customer already runs - camera streams, document stores, identity providers, and finance systems - so a deployment reads from the record of truth instead of asking for a parallel copy of it.
Azure-first infrastructure
The cloud path is built on Azure first, so a private tenant deployment lands in a subscription most regulated customers already hold, under governance their own cloud team already operates.
The architecture
The platform, drawn
Products on top, shared layers in the middle, deployment surfaces at the base. Read it downward: a product never talks to the hardware directly, and the boundary at the bottom is the one your data stays inside.
Ownership
Built and owned by FalcRise
FalcRise designs, builds, and owns all four products outright. Every model, pipeline, and line of application code is our own - nothing is resold, white-labelled, or licensed from a third party.
FAQ
Questions about the platform
What does self-hosted AI software mean?
Self-hosted AI software runs the model and the data it processes inside infrastructure the customer owns or controls - a server room, a private cloud subscription, or an edge node on site. Nothing is sent to an external API for processing. Every FalcRise product ships this way, because for regulated customers it is the condition of adoption rather than a preference.
What is an on-premise AI deployment?
An on-premise AI deployment installs the software on hardware inside the customer's own building. Inference, storage, and audit logs stay behind their network boundary. The FalcRise platform treats on-premise as a first-class surface: the same release that runs in a private cloud tenant runs on-premise, with no reduced version.
Does FalcRise own the software it deploys?
Yes. FalcRise designs, builds, and owns all four products outright. Every model, pipeline, and line of application code is our own - nothing is resold, white-labelled, or licensed from a third party.
Can one platform run computer vision and a private LLM?
Yes. Computer vision and language model serving are both inference workloads on GPU hardware, and on the FalcRise platform they share the same inference layer, the same access control, and the same audit record. That is why VMS + Omniscient and OrgBrain deploy the same way despite doing very different jobs.
How long does a deployment take?
The platform deploys in weeks, because the deployment is a software installation into an existing environment rather than a build starting from nothing. The products already exist; a deployment configures them against the customer's cameras, records, and access rules.
Which cloud does the platform run on?
Azure first, for private tenant deployments. The platform is not tied to a single provider at the infrastructure level, but Azure is the path that is built, tested, and shipped, because it is the subscription most regulated customers already hold.
See the platform running.
Pick the product closest to your problem and we will show you the deployment behind it.