Work
A cold-chain command centre
A working prototype: ask in plain language what is happening across a temperature-controlled distribution network and see excursions, delays and stockout risk on a map.
- What it proves
- AI over a physical supply chain where data integrity matters.
- Built from
- Natural-language querying of operational state
- Temperature-excursion detection
- Predictive stockout signals
- Predictive delivery-delay signals
- Geospatial map of the distribution network
The problem
A temperature-controlled distribution network produces more operational data than a dashboard can show: shipment positions, temperature logs, delivery windows and stock levels, all changing at once. A cold-chain breach carries different consequences from a late delivery. This prototype replaces another dashboard with questions in plain language.
What we built
An operator asks a direct question, for example which shipments risk a temperature excursion in the next six hours. The system chooses which data sources answer it. A map shows where a developing problem is and how far it reaches.
The system watches for temperature excursions as they develop. It also raises predictive signals for stockouts and delivery delays before they become incidents. The same state that answers a typed question drives the alert the operator sees first.
A reading that is wrong, late or missing can mean missing an excursion, so each answer carries the provenance of its readings. A reply separates a confirmed excursion from a sensor gap that only looks like one, and it shows uncertainty instead of smoothing it away.
What it proves for a client
Natural-language access to physical supply-chain state can respect integrity rules, with provenance carried in each answer. Pharma and logistics both need to show where a number came from.
Method stages
It sits at Stage 02 · Prove the value. Users on one bounded part of a network would come in Stage 04 · Pilot in one place, after the Stage 03 · Engineer the system work. See the KARPINITY Method.