ZillA xZane

Factory and warehouse automation

AI on the line and in the warehouse, with the machine's own safety still in charge.

Sorting cells, mobile robots, conveyors, sorters and palletisers already have controllers and safety systems, on the production line and in the warehouse. xZane governs what the AI may ask of them and records every command, rejection and stop. You write the adapter once and reuse it on the next line.

Start a pilot conversation See the command path
Runtime responseinterrupted
// Legal command, stopped by the world
{
  "status": "interrupted",
  "outcome": "interrupted_by_reflex",
  "reason": "reflex_latched_mid_motion",
  "latch_reason": "front",
  "ai_command": null,
  "confidence": null
}
A real response from a ROS 2 robot. A valid command was stopped mid-motion when the robot's own safety latched on an obstacle in front.

The layer you rebuild

The robot is the smaller part of the cell.

25–40%
of cell cost is the robot hardware1
30–50%
of a project goes on integration2
8–26 wk
for a custom robot integration, from a simple cell to a full custom cell3

Add AI and the integration layer grows: command checking, a capability schema, hardware abstraction, safety gating and traceable logging. Our estimate for that layer alone is 4 to 6 weeks of engineering per project. It does not bill as integration work, and you build it again on the next project.

With xZane the adapter is written once for a hardware model. The second and the hundredth robot of that type need very little additional integration time.

In practice

Where integrators use it.

Mobile robots

AMRs that move pallets, totes and parcels, in a plant or a warehouse. ROS 2 is the stack xZane already runs on, with speed, duration and direction limits held in the schema.

Sorting and picking

Picking cells and parcel-sorting stations. Perception and mapping run as self-contained modules. YOLO detection and SLAM modules exist today and deploy without regression-testing the rest of your stack.

Conveyors, sorters and PLC equipment

Expose only the intents you want the AI to have, such as start, stop and speed within a band, through an OPC UA or Modbus adapter.

Palletising and packing

Drives and palletisers behind a Modbus or vendor adapter. The AI proposes, the schema bounds the command, and the machine's safety circuit stays in place.

Swap the model, keep the cell. When a client changes from Claude to a local Qwen or Mistral, the new model targets the same capability schemas. The hardware does not know the brain changed.

Safety

xZane puts the device’s built-in safety above the AI.

When a machine's own safety trips, it tells xZane, and xZane refuses AI commands to that machine until the safety clears. The AI can ask for a reset but cannot force one. The rest of the line keeps running.

What xZane does not do. xZane is not a safety-rated controller. It does not replace certified safety functions such as safety PLCs, safety lidar or emergency stops. It keeps the AI inside the limits you declare and obeys the safety your machine already has.

The command path

From a model's request to a machine.

Documentation

Inputs for the cell's risk assessment.

A robot cell is machinery. The Machinery Regulation (EU) 2023/1230 applies from 20 January 20274 and expects a documented risk assessment of the integrated cell, for industrial robots under ISO 10218-2, together with traceable technical documentation. That assessment stays with you as the integrator. Driverless industrial trucks, which includes many AMRs, also have their own safety standard, ISO 3691-4.

xZane gives you two inputs to it: the declared envelope, which is the capability schema, and a ledger of every command, rejection and safety stop. Where a system falls within the scope of the EU AI Act, the same ledger supports its record-keeping requirement for high-risk systems (Article 12).

Pilot

One machine, one use case.

A bounded sprint: one device, one use case, four to six weeks. We write the adapter together. At the end xZane is running live on your hardware and your engineers understand every part of it.

Standard pilot

€10,000

Single machine · standard hardware · 4 weeks

  • xZane running live on your machine
  • An adapter written with us and owned by your engineers
  • A ledger of every command, rejection and safety stop
  • SDK access, so you can write future adapters without us
  • 2 to 3 hours a week of an engineer's time

Extended pilot

€25,000

Multi-device cell or custom hardware · 6 weeks

  • Everything in the standard pilot
  • Multi-device cell integration
  • Adapter development for less common hardware
  • Ground Control UI, so operators watch AI decisions in real time

Ongoing licence: €400 to €600 per device per year. A 10-device cell runs at €4,000 to €6,000 a year. xZane installs on a central hub, and devices connect through adapters.

Start a conversation

A described machine is enough to start.

Tell us about the machine and what you want a pilot to prove. A 30-minute call is usually enough to scope it. No commitment required.

Open the enquiry form ↗

Or write to [email protected]

What we need from you

  • One machine or test cell
  • A use case: what you want it to do through xZane
  • 2 to 3 hours a week of an engineer's time
Sources
  1. Robot hardware is typically 25–40% of total cell cost. AMD Machines, True Cost of a Robot: 5-Phase TCO Breakdown; Zeueeauto, Robot Integration Guide.
  2. Integration routinely runs 30–50% of total project cost. AMD Machines (as above); Cost of deploying robotics on a factory floor (2026).
  3. Custom robot integrations commonly take 8–12 weeks, full custom cells 12–26 weeks. Zeueeauto, Robot Integration Guide.
  4. Machinery Regulation (EU) 2023/1230 applies from 20 January 2027. EU-OSHA; Pilz.