The platform
MarshallAI is a node-based deep learning software built for industrial environments. Every stage of the vision process — camera input, preprocessing, model inference, and output — is a connectable node that operators wire together and reconfigure without writing code.
Deployment
The platform runs on any compatible on-premise hardware or our robust REID-4 edge device. MarshallAI integrates directly with your existing production systems through native industrial protocols.
No cloud connection is required at any stage. Your process data, models and detections never leave the facility.
Any input.
Any standard.
One unified pipeline
MarshallAI’s multimodal pipeline handles all input signals inside the same node graph, on the same device, in the same inference cycle. The output possibilities are virtually limitless. A typical output can be a dashboard, database, PLC, SCADA or rejection system.
Built for industrial grade intelligence
A fully customizable
processing path
Each detection program is composed as a graph. Nodes represent discrete operations — eg. a camera source, a resize transform, a neural network inference block, a threshold filter, an OPC UA write.
You connect nodes in the order your process requires, set the parameters for each node, and the platform executes the graph in real time on every incoming frame.
When a product or process changes, you simply modify the relevant nodes. This means new SKUs can be added in minutes.
Node-Based Vision Pipeline
