Every case study here covers a venture we own. That is deliberate rather than convenient: client engagements carry confidentiality, so the only work we can document at the level of architecture decisions is our own.
Client case studies publish only with written permission, which is the same commitment set out in our confidentiality terms. We would rather show our own decisions in full than someone else's in outline.
Each study covers one venture from context through to the decisions we would defend in a review. Where a decision has since looked wrong, it stays in the record.
Most case studies are sales collateral with a graph attached. These are written for an engineer deciding whether our judgement is worth buying, which means the rejected options matter as much as the shipped ones.
What existed before, who it was failing, and why the obvious solution had not already been built. Written so someone outside the sector can follow it.
Not a feature wish-list. The specific technical and commercial constraints that made the work hard, stated plainly enough that you could disagree with our reading of them.
Every material call includes the alternative we turned down and why. A decision without a rejected alternative is not a decision, it is a preference.
Qualitative where the number does not exist yet, specific where it does. We publish market figures with their source and never attach a percentage improvement we cannot evidence.
Articles cover narrower technical ground. The venture index covers the products themselves. The doctrine explains why the decisions above went the way they did.
Notes on architecture, applied AI, cloud operations, and delivery discipline.
Aurelia, Voyazio, and AxleOne — stage, sector, model, and availability.
The eight clauses that governed every decision in these studies.
Decision records are a deliverable on every engagement, not a marketing artefact produced afterwards. Send the brief and the first one arrives with the architecture note.
