Condition monitoring
Vibration, temperature and current watched on critical assets, continuously.
Predictive maintenance
Remaining useful life estimated per component, with work orders raised before failure.
Spares and work orders
Parts and technicians planned against forecast failures, not against the last breakdown.
Root cause across shifts
Stops traced to the variable that moved, across shifts, batches and suppliers.
Industries/Physical economy
Gallahand forManufacturing and Industrial Plant
An error rate that is fine in a chat window is catastrophic on a line. We build verification first, then autonomy, into plants that cannot afford to stop.
Measured in: cost per unit · unplanned downtime
Line 2 · Vision QAStation 4 · 1,820 units/h
OEE 84.2%Escapes 0
INSPECT · unit 40,112 pass
The constraint
Downtime is the only cost that compounds by the minute.
Downtime
An unplanned stop costs output, labour, energy and the customer's trust at once.
Quality
Escapes found by the customer cost more than everything spent on inspection.
Energy
Energy is now a variable cost per unit, and most plants cannot measure it per unit.
Use cases
What we build for manufacturing.
Each use case runs on the same stack, so the second one costs a fraction of the first.
Vision inspection
Defects detected at line speed, with the image, the model version and the decision stored.
Process drift detection
Parameters watched for drift before they produce scrap.
Batch genealogy
Every unit traced to its materials, settings and operators.
Supplier quality
Incoming quality scored per supplier and fed back into purchasing.
Energy per unit
Consumption allocated to lines, products and batches in real time.
Load shifting
Energy-intensive steps scheduled against tariff without missing a delivery.
Line scheduling
Changeovers and sequences optimised for throughput and energy together.
Private models on plant compute
GRAIL runs inside the plant perimeter, so production data never leaves the site.
Integration
We read from existing PLCs, historians and MES through standard industrial protocols. Models run on plant compute where the data must stay inside.
Platforms in play
One stack underneath.
CORBENICLayer 02 · Bits
The operating picture. Telemetry, state and memory joined into one live model of every asset.
GRAILLayer 01 · IntelligenceThe intelligence core. Model routing, agents, evaluation and the policy envelope, on your compute or ours.
CALIBURNLayer 04 · ElectronsCharge, dispatch and energy on one schedule, against tariff, duty cycle and the next shift.
How an engagement runs
One number, moved and proven.
01 · SURVEY
Find the constraint
Two to four weeks on the asset. The number that costs most, the data that already exists, and what has to work first.
02 · INSTRUMENT
Connect what runs
Existing systems connected and gaps closed with vendor-neutral hardware. A baseline both sides sign.
03 · PROVE
Move one number
A bounded deployment against the baseline, reported in physical units, with a human gate on anything irreversible.
04 · OPERATE
Run it, or hand it over
We operate the system with your team, or transfer it with the code, the models and the playbook.
We do not publish client names or case studies. References are shared in a briefing, under agreement.
From the thinkingThe Tolerance WarInside the plant, the contest is over error rates consumer AI was never built to meet.Read ↗Briefings
Send the constraint, not a brief.
Most of our work arrives by introduction. If you run or back a manufacturing operation, tell us the number that is costing you most.
- 01The plant, the lines and the product.
- 02The number costing you most: downtime, scrap, energy or throughput.
- 03Control and data systems: PLC, historian, MES.
- 04Where data is allowed to go, and where it is not.