Business cases
The problem, then a path.
Each brief is named by the problem, not by a client. It points at products and services we actually offer, and at a Work study only when one exists.
A conversation on top of the system, not instead of it
The organisation wanted a system. It bought a reply box.
A general agent is not an organisational system
A capable general agent still leaves the organisation without a system it can run.
Ask the question of your data
The numbers moved. Nobody can say what happened, for whom, or what to do next.
Capability inside the organisation
People are fluent about models and still cannot judge a design they will have to live with.
Find the companies that need you
The team has a list, a quota, and no honest account of why this company, now.
From complexity to a course of action
The organisation can see the problem. It cannot yet see a bounded way through it.
From experimentation to operating practice
AI remains a series of experiments with no owner, no standard, and no path into the work.
One customer, one conversation
The customer has already explained the problem. The next channel does not know.
Search a hard problem, do not declare it solved
The question is precise, expensive to explore by hand, and too easy for a fluent model to fake.
See what users do, then investigate why
Activation fell. The dashboard cannot say where, for whom, or what changed just before.
Strategy under uncertainty
A single narrative about competitors, customers or regulators is being treated as a plan.
Taking AI into production
The organisation has a prototype that cannot yet leave the room it was built in.
When a product is not the whole answer
The workflow is real, and no product already owns its interfaces.
Why an operating system, not a chatbot
Teams buy a conversational interface and later discover they needed a runtime.
Work out what to build — before you build it
There is pressure to deploy, and no shared account of what a good outcome would be.