Intelligence
PLJ ONE combines signals from operations, finance, customers and growth — without turning your business into a black box.
The design principle: an answer should arrive with the records that produced it, how sure the system was, and a way to check and correct it. That is what we are building towards.
“Why were expenses higher this month?”
Delivery fees, not ingredients. Up 21% on three small runs.
Intelligence
It should live inside your business. Each product becomes more useful the more of the business the ecosystem understands.
Product A sales up 31% this week
Demand is real, and it is accelerating.
DemandMargin on Product A holding at 64%
The demand is worth serving — it is not being bought at a loss.
MarginDemand plus margin supports a stronger push
Spend where the business already wins, not where it hopes to.
Timing142 previous buyers have not returned this month
The best audience for that push is already in the business.
AudienceOne worked example
Nothing is hidden between the records and the recommendation, and the last word is yours.
Lunch demand is rising and margin is holding. Increase Friday lunch prep modestly — there is no case for discounting into demand you already have.
An illustrative example, not a customer or a guaranteed outcome. Figures are made up to show the shape of the reasoning. Cost signals are read from documents in PLJ Account today — the till-to-books connection is ready but not yet live.
The rules we hold ourselves to
Inference reads a receipt and proposes what it says. It never owns the money maths, permissions or a tax conclusion. Totals are exact integer arithmetic, every time, because a number that is 99% right is wrong.
When PLJ Account reads a document, each extracted field is stored with how sure the system was. Low confidence surfaces as one specific question, not a blank form to fill in again.
A receipt or invoice is filed alongside the figures read out of it, rather than replaced by them, so an answer can be traced back to the document behind it. An answer you cannot trace back to a document is an opinion.
“Why were expenses higher this month?” is answered from the business's own records, with the transactions that produced the answer attached — so the reply can be checked rather than believed.
What is read off a document is a proposal, not a decision. The figures, dates and categories are shown for you to check, and anything read incorrectly can be corrected before it becomes a record.
Cross-product signals exist to help one business understand itself. They are not a pool to be mined across customers, and nothing leaves a business to make someone else's product smarter.
Three honest stages, so nobody has to guess which part of the example above is already running.
PLJ POS surfaces sales, stock and margin signals inside operations. PLJ Account reads documents, categorises spend and answers questions about the money — each within its own product.
Signals cross the boundary: cost from PLJ Account informing margin in PLJ POS, and sales from PLJ POS informing the books without re-entry.
One shared picture that every PLJ product reads from — the point at which a recommendation can weigh demand, cost, customer and campaign together. This is the direction, not a shipping date.