Full Case Record

Three cases, published in full.

Different sectors, one mechanism: turn work that was already happening into a record that holds up in front of a third party. Client identities are withheld. Every figure published here carries the assumptions it rests on and the gaps that were not measured, exactly as the source documents state them.

Each case is published with its working attached: the figures sit beside the assumptions they rest on and the gaps that were not measured, so the arithmetic can be checked rather than taken on trust. Read them as a record of method — what was deployed, what it returned, and what remains open — not as a results summary.

The three cases

Rows of fruit trees in an orchard at first light, photographed from the lane between them
SectorFruit export · South America

When the field visit became the contract

An export operation ran its agronomic visits on paper, chat and spreadsheets. Digitising the visit was expected to save time. What it actually produced was a system of record: the platform became the place where export volume was committed, supplier certifications were vetted and compliance was evidenced — not a field notebook, but the official record behind the contract.

field visits on record
1,365
report generation rate
83.6%
prospect-to-contract ratio
199 → 100

The engine that turns a field visit into a defensible record of a volume commitment is the engine that turns an RFI into a retrievable claim file on an EPC program.

Client identity withheld. Method, figure set, assumptions and open gaps published in full; no figure is rounded.

Read the full case →When the field visit became the contract
Empty supermarket aisle at night under fluorescent lighting, product labels unreadable
SectorMulti-site retail · Northern Europe

From detection to execution, and to cash

An in-aisle robot already detected the problems on the shop floor every night. Nobody was resolving them: the alerts arrived without order, without an owner and without a value attached. The work here was the layer that closes that circuit — every alert becomes a task with a price, a rank and a named owner, and a record of what was actually done.

assigned tasks resolved
317 of 318
value of a selected task vs. a random alert
3.1x
of exposed value captured from 3% of alerts
9.4%

The engine that ranks nightly shop-floor alerts by the money each one recovers is the engine that ranks an open RFI backlog by the money each item exposes on an EPC program.

Client identity and robotics vendor withheld. Method, figure set, assumptions and open gaps published in full; no figure is rounded.

Read the full case →From detection to execution, and to cash
Stacked wooden crates in an agricultural packing shed at dusk, no people or branding visible
SectorAgro-industrial export

From blind spend to an Operational Brain

Spend executed remotely by a third party in the field, reported 30 to 45 days after the fact, on an operation whose income concentrates into a few weeks of harvest. Three stages, each paying for itself before enabling the next: capture the data, automate the flow, then a read-only layer over the systems already in place that ranks margin leaks by money.

budget-vs-actual latency
30 → 0 days
released from manual data entry
~110 h/month
of purchase spend at risk as VAT credit
~19%

The engine that makes remote third-party spend traceable to a cost centre on the day it occurs is the engine that makes spend traceable to the unit of work that originated it on an EPC program.

Client identity withheld. Method, figure set, assumptions and open gaps published in full; leak-ledger amounts are in the local currency of the case, unconverted.

Read the full case →From blind spend to an Operational Brain

← Back to Documented Case Contexts