We did not set out to invent a new technology category. We needed a Retriever.
At The Old Dog, the tools did their individual jobs well enough. One knew about purchasing, another held recipes, stock lived somewhere else, sales arrived from the EPOS system and finance had its own sensible view of the world.
The difficulty was not a lack of information. It was that none of it arrived together when somebody needed to make a decision. The systems could each provide part of an answer, but the person running the venue still had to do the joining up. That is why Ross asked me to build Retriever.
We began with a working pub
We began with a working pub in Derbyshire, not a whiteboard full of technology categories. The aim was practical: connect purchasing, recipes, stock and service so people could see what mattered and decide what to do next.
Only later did we find the phrase for it: decision layer technology.
The layer between records and action
A decision layer sits between the systems that record what has happened and the people accountable for what happens next. It brings together the relevant evidence from the systems already in use, adds the context needed to understand it and turns it into a clear decision for someone to consider.
The EPOS system remains the source for sales, the supplier invoice remains the source for what was bought and the recipe remains the source for what ought to have been used. Nobody needs another system claiming to be the official version of everything.
Context is the important bit
A stock number is not very useful on its own. Is it above or below par? Has demand changed? Is there an order already on the way? Is the item available from the usual supplier? Is the variance likely to be a counting issue, a recipe issue or something more interesting?
A useful decision layer turns those loose pieces into something practical: review this invoice, prepare this order, check this recipe or ask this question. It gives the person running the venue a better starting point for judgement.
Four jobs, one human decision
A useful decision layer does four things. It retrieves the evidence that belongs to the question. It adds the operational context that gives those records meaning. It frames the decision clearly enough for someone to consider. Finally, it shows what it found, states its confidence and makes uncertainty visible.
The person with the responsibility keeps the authority. Giving an organisation more information does not necessarily improve its judgement; it can simply allow the same poor decision to be made with a more impressive chart beside it.
Why the name Retriever fits
A retriever goes out, finds the relevant thing and brings it back to the person who needs it. It does not invent the rules or take responsibility for the outcome.
That is the behaviour we wanted from the technology: go into the operational detail, connect the clues, return a useful picture and help an owner or manager make the next move without asking them to become a systems integrator before breakfast.
The same principle can travel
The same principle can apply in a larger organisation, where a decision may sit between procurement, finance, operations, supply chain, HR and legal. Each function can hold a legitimate part of the answer while the organisation remains slow because the decision exists in the gap between them.
A decision layer can bring those views together around the actual question, show the evidence and assumptions, identify the people who need to decide and leave behind an honest record of why the decision was made. The underlying systems remain the sources of truth; the accountable people retain the authority.
We did not build this because the world needed another piece of software with an ambitious noun attached to it. We built it because the gap between information and action is where much of the real work happens.
Decision layer technology is our attempt to make that gap smaller, clearer and more useful. Technology earns its place when it leaves people better able to decide.