Case studies

The evidence behind the work

Real engagements across financial services, energy, education and industry. Each number states where it comes from — measured, client-reported or modelled.

Professional services — our own firm

Governed agent operations, tried first on our own back office

Agents & automationGovernanceSoftware

12 stagesValidation pipeline between an agent's proposal and the record

Global payments fintech

An applied-AI sprint for a People function, set against its plan

AcademyDiscovery

~5 weeksFrom first workshop to first client-confirmed working workflows

Multilateral development bank

Work in delivery

Evidence before automation: discovery for a development bank

DiscoveryGovernance

5AI solutions to prioritise, each with a full dossier

Global payments fintech

Payroll consolidation as four controlled steps, demonstrated end to end

Agents & automationSoftware

4 stepsPrepare, validate, consolidate, review: shown end to end

Multilateral development bank

Solution design

A seven-line transformation programme, prototyped before any contract

ArchitectureSoftwareAgents & automation

7candidate workstreams proposed across the lending lifecycle

Regulated retail bank

A role-based programme to activate a retail bank's idle AI licences

AcademyGovernance

6Use-case families covered, from document analysis to agents

Global payments fintech

People analytics and review-cycle AI: what stuck, what stayed partial

AcademyAgents & automation

3people systems in the API data work; one confirmed, two partial

Private university

Solution design

An AI study companion, costed per student before any build

Agents & automationSoftwareGovernance

US$500Designed hard cap on model spend per cohort of ~3,000

Upstream energy operator

Solution design

Designing a field command layer that assists and never authorises

ArchitectureAgents & automationSoftware

7operational stages mapped in one prototype command view

Multilateral development bank

Solution design

Designing a pre-review agent a sceptical accountant can check

Agents & automationGovernance

~40,000Pages a year the design is sized to process

Commercial bank · nationwide branch network

Solution design

Designing an AI operating model for a nationwide commercial bank

DiscoveryArchitectureAcademy

3Dimensions in the designed model: skills, process, joint build

Creative and media production agency

Solution design

An AI-assisted campaign pipeline, designed around the human decision

Agents & automationAcademy

2scripted demonstrations built for a proposal, on fictional data

Postgraduate business education

An agentic tutor and a two-agent debate for postgraduate teaching

2 artefactsBuilt for teaching: a tutor agent and a two-agent debate demo

How to read these cases

Every number carries its source

We would rather publish a modest figure we can defend than an impressive one we cannot. Three labels travel with every metric on this site.

Measured
Counted from the engagement's own delivery records, systems or logs.
Client-reported
Stated by the client organisation. We attribute it; we do not audit it.
Modelled
Projected from the engagement's business case or baseline. The note says what the model rests on.

How far the work went

Delivered work
The work described was carried out and handed over. Each case says how far it went.
Work in delivery
Work under way. Outcomes are reported as they are verified.
Solution design
We designed or prototyped a solution. Figures are scope or modelled; nothing is claimed as deployed.

Logos on our home page are not a key to these cases. They mark where our team has delivered work, not which organisation a case describes.