Payments and branches
Authorization is already on a short clock. Score the transaction on the server that holds it, before cardholder data leaves the network that is allowed to see it.
Edge AI
for Enterprise
HALO runs the model on the branch server, the payment terminal, the vehicle gateway, and the air-gapped node. The data stays where the decision is made.
JavaONNXCPUAir-gappedSelf-hosted
Who it is for
The people who already run the edge.
Fleet and telematics
Driver and vehicle signals are scored in the cab. A summary comes back when the link is up. The raw stream does not ride the cellular bill.
Sovereign systems
The node keeps working with the network unplugged. The package is small enough to review, and the runtime does not fetch a model or a library from outside.
On the node
A workload, not another platform to operate.
- One package JVM and ONNX in a runtime sized for a branch server or a gateway, on the CPU you already have. A GPU or an NPU is optional.
- Scored in place Tabular models and vision run next to the event. A cloud round trip is no longer part of the decision.
- Explained where it ran A decision can carry a local explanation. The audit stays with the event, on the node.
- Still running when the link is down Inference continues. The log waits, encrypted, until the network returns. A new model can replace the old one without restarting the process.
A 30-day pilot
A few of your nodes, your models, your network.
The operating system, the firmware, and the disk stay with what you already have. HALO is the inference runtime on top. We package the models you already have, put the runtime next to the traffic, and measure a live score, a local explanation, a model swap, and a run with the network unplugged. Tell us where the nodes sit and what you score today.