Worked examples
Ten worked examples, each from a different sector, each turning on a different kind of decision. They exist to show how a decision gets made: the evidence that moved it, the options rejected, and what it would take to change the answer.
Illustrative. These are illustrative worked examples, not exact engagements. They are written to show the method rather than to claim results. Figures taken from published sources are cited; everything else is labelled as an assumption, exactly as it would be in a real Decision Pack.
Maritime logistics
A terminal operator wanted to predict vessel arrival times to plan berths, tugs and labour. The data the model would learn from turned out not to exist in usable form, so the money went to the contract that produces the data instead.
Evidence stopped a project that every stakeholder wanted.
Veterinary care
A practice group wanted an AI assistant to advise owners out of hours. Professional rules on what constitutes an animal being “under care” meant the assistant could never do the valuable-sounding part, so the design changed to the part it could legitimately do.
A regulatory rule turned a broad product idea into a narrow, useful one.
Specialty insurance
A marine cargo underwriting team wanted to automate submission intake. Treated as one decision it was a stalemate; split into two, document handling and risk judgement, it answered itself.
Build versus buy answered per component, not per project.
Waste and recycling
A materials recovery facility was ready to buy robotic sorters to hit an export contamination spec. Measuring the line showed the machines it already owned were not the limit. The incoming material was.
Capital spending aimed at the wrong constraint.
Live events and ticketing
A ticketing platform wanted an ML bot-detector. Every control that replaced it is standard anti-bot practice and none of it is proprietary; what the engagement decided was to refuse the classifier, and to start measuring whether any of it works.
The controls were off-the-shelf. The decision was refusing the model and finally measuring the thing everyone argued about.
Agriculture
A cooperative wanted a predictive model for grain spoilage across member silos. The measurement that detects spoilage earliest is a cheap sensor and a threshold. The model would have been an expensive way to be later.
A prediction model that a sensor and a threshold made unnecessary.
Legal services
A litigation team wanted a large language model to cut a disclosure review. The binding requirement was not accuracy but the ability to explain and reproduce the cut to a court. That ruled the model out of the decision and into a narrow assisting role.
A non-functional requirement decided which technology was allowed.
Pharmaceutical logistics
A distributor drowning in temperature-excursion investigations wanted them automated. The regulation settles quickly who signs. The hard part was the failure nobody asks about, an assistant good enough to be approved without being read.
A fluent draft is persuasive whether or not it is right, so the edit rate became the thing to watch.
Public transport
A bus operator asked for machine learning to improve driver rostering. The algorithm question had a settled answer, because hard legal limits want a constraint solver. The finding worth the engagement was the platform the previous team had left behind, costing more each month than the problem was worth.
Two decisions kept deliberately apart. One settled by the literature, one nobody had asked about.
Cultural heritage
A museum wanted to clear a cataloguing backlog with machine-generated metadata. The technology worked; the risk was that a wrong guess written into the catalogue of record is effectively permanent. The architecture followed from that.
A decision driven entirely by how hard it would be to undo.
When an engagement completes and a client agrees to be referenced, its write-up appears here in the same structure, with measured outcomes, and marked as a real engagement.
Problem
The friction, stated in the client's terms.
Context
Organisation, product, users and objective.
Constraints
Hard limits and strong preferences, with their source.
Current Architecture
What existed, as a fact, before any change.
Evidence
Each statement placed on the ladder: fact, assumption, hypothesis, constraint.
Questions that changed the answer
The ones that reordered the options.
Options
Every direction considered, including doing nothing.
Economics
Build, run, change and exit cost: the four horizons.
Decision
What was chosen, why, and why the alternatives were not.
Trade-offs & reversibility
What is accepted, and how hard it would be to undo.
Complexity budget
Which elements earned their place, and which did not. Present where it applied.
Non-functional requirements
What the system must achieve, stated as requirements.
Decision Gate
GO, PAUSE or STOP, with what would change it.
Implementation
What was actually built or changed.
Outcome, or expectation
Measured results on a real engagement; on an illustrative one, what the decision was expected to achieve and how it would be proven.
Lessons
What transfers to a different problem.
Free · 30 minutes · one real problem
Thirty minutes, one real problem, structured thinking. If there's no value, there's no engagement.
