Anonymized summaries of client engagements: what was needed, what we built, and what changed. Each case study is paired with the method used.
We publish our work in two forms. The use-case library below documents systems we run in our own multi-entity family business, written as anonymised patterns: the shape of the problem, what was built, and what it required. These are live now. Named client case studies are published separately, and only after the client has approved the anonymised version — the first of those are noted at the foot of this page.
Every published item is anonymised. What we keep is the part that transfers to another business; what we remove is anything that identifies a client or that no reader outside the engagement should see. We do not publish invented clients or invented numbers.
Each links to a full write-up. All are systems in daily use inside our own companies, published as transferable patterns rather than named accounts.
Every work item becomes preserved, classified, verified, machine-readable evidence. The document is the source of truth; humans supervise, bounded AI does the work.
Read the case →People, policies, projects, decisions, and precedents held as owned, dated, linked entries — so the answer is found, not remembered, and survives staff turnover.
Read the case →A raw, multi-topic voice note becomes routed, structured output — actions with named owners, in the format each recipient needs.
Read the case →One morning brief covering deadlines, decisions waiting, and the cross-entity signals no single inbox or tracker shows.
Read the case →Inbound mail read, classified by what it needs, and routed — with a surface showing what is waiting, what is approved, and who is waiting on whom. Replies drafted, never sent.
Read the case →A large structured information request — due diligence, audit, tender — turned into tracked items, each answered only from the canonical source, with gaps flagged rather than guessed.
Read the case →Entity briefings, cross-entity signals, and overnight work products produced on a fixed daily schedule — agents produce, the operator approves and acts.
Read the case →See the full use-case library, organised by business function, or the methodology library for the practices behind the work.
A named study is published only after the client has read the anonymised version and agreed to it. Several are written and none has been published, because in a market this small a pseudonym often is not anonymity — a specific enough description identifies a company to exactly the people who would recognise it. We would rather hold a case than publish one a client did not expect to read about themselves.
What transfers out of an engagement is the method, not the anecdote. Each practice in the methodology library describes one named approach — when it fits, how it works, why it works, a worked example, and the ways it goes wrong.
Read the practices →The same instruments we use inside engagements, published in full: a dependency map you can fill in tonight, the four questions we open with, and the diagnostic to run before anyone sells you a system. No sign-up, nothing collected.
Open the tools →Each anonymized case study is approved by the client before publication. Until then, the methodology library describes our work.