Built for businesses with more data than they can use
Big enough to have a real data problem, not big enough to interest the global consultancies. That is the gap we exist to fill.
You will recognise yourself if
Your systems record how a transaction was coded. The document itself, the invoice, the contract, the delivery note, holds the detail that actually matters, and it is effectively invisible.
Your best decisions depend on one or two experienced people and what they remember
You have looked at AI, agreed it matters, and not found anyone credible to have the conversation with
You have years of operational history that nobody can query
Month end reveals things that should have been caught weeks earlier
Two or three of these is usually enough to be worth a conversation.
WHERE WE GO DEEPEST
Three sectors where our references and our questions are sharpest
We work outside these three. They are simply where our references and our questions are sharpest.
High transaction volumes, thin margins, supplier pricing that drifts, stock in the wrong place. Small changes in visibility move real money.
-
Supplier price drift
-
Invoice volume
-
Stock placement
-
Thin margins
Job costing that only tells the truth at the end, subcontractor and supplier detail buried in documents, and margin leakage that shows up too late to fix.
-
Job costing
-
Subcontractor detail
-
Document sprawl
-
Margin leakage
Demand that is hard to forecast, inputs whose prices move constantly, and production knowledge that lives in the heads of the people who have been there longest.
-
Demand forecasting
-
Input price volatility
-
Undocumented knowledge
EXAMPLE USE CASES
What the work actually looks like
Two examples, drawn from work under way. Both are told as the work and the decisions rather than the results, because the interesting part is what changed along the way.Thousands of supplier invoices a month, processed by a small accounts team under constant month-end pressure. Invoices get captured, coded, approved and paid. The step that would catch an overcharge, a rate that has quietly crept up, or a part bought cheaper elsewhere in the group does not happen, because there is no time and no view across it.
The deeper issue: every decision was being made on how an invoice had been coded, not on what was actually on it.
A Discovery Workshop, then a proof of concept against several years of historical extraction data, loaded into a segmented cloud environment and queried in plain language.
The proof of concept did two things. It showed that the analysis the business wanted was possible and useful. It also exposed, unambiguously, that the existing extracted data was not reliable enough to make decisions on, with duplicates, unverified fields and missing links back to the finance system.
Rather than build on a foundation we knew was weak, the next phase was rescoped to establish a verified one: re-reading the source documents with independent verification, linking every line back to the coding and master data, and keeping it current so the gains hold rather than decay.
Inventory is the largest asset on the balance sheet and it turns roughly once a year. Reordering runs on printed reports and long experience. Stores request transfers by email. Stock that moves for a customer viewing tends to stay where it landed, because sending it home is an admin job nobody is rewarded for.
Nobody has a systematic view across the network.
After the Discovery Workshop and a series of working sessions, inventory intelligence was confirmed as the highest-value use case. We adapted a demand forecasting framework that has been in production use for a decade, loaded the business's own sales, stock and transfer history, and encoded the buying rules that until then existed only in one person's head, written by her, in plain language.
The original plan was to move straight from workshop to a production build. Partway through, with real questions still open about lead times, outstanding orders and the reliability of an upstream system, we recommended against it.
Instead we deployed the working prototype online for the team to use on real weekly reorder and redistribution decisions, comparing its recommendations against what the buyers would have done anyway.
BEING STRAIGHT WITH YOU
Who we are not for
We would rather say this up front than waste your time.
Very small businesses
If you run one system with clean data, the AI already built into it is probably good enough. Save your money.
Large enterprises
Above a certain scale the global firms will serve you better than we can, and they are set up for it.
Anyone wanting AI for its own sake
If there is no business outcome attached, we are the wrong firm.
Start with a conversation, not a commitment
No preparation, no slides, no obligation. Tell us what is frustrating you and we will tell you honestly whether there is anything here worth doing, and what the first step would cost if there is. If the answer is that you do not need us yet, we will say that too.