Which AI idea is ready for a pilot?
A finance team, a service desk and a facilities team each bring a proposed pilot. Open their working papers to see how the owner, samples and existing process change the recommendation.
Scope my AI pilotInteractive example · fictional records
Open a candidate brief
Pilot shortlist · Operations review
Redacted invoices and receipts
Policy answersSample materialTwo current-looking policy editions
Facilities dispatchSample materialTickets with category and site
- Sample material
- Tickets with category and site
- Decision owner
- Duty dispatcher
- Starting baseline
- Routing table already exists
Try the dispatch rule on past tickets
Most sample assignments follow a site and category table. Replay that rule and inspect the exceptions before deciding whether interpreting free-text descriptions merits an AI pilot.
Scope my AI pilotWhat we can build with you
A pilot charter with one accountable owner, a workable sample, a comparison baseline and a decision to proceed, revise or stop.
- Bring to discovery
- Bring three candidate processes, a person who runs each one and a sample of the difficult cases.
- How to test the pilot
- Agree what a wrong decision would cost. Replay reviewed examples alongside the current process and use those results to choose the first deployment.
Implementation details and questions
Business, technology and product leaders choosing where to invest and how to turn a priority into a deliverable system.
Turn a business ambition into a delivery decision. We assess workflows, product opportunities, existing systems and AI readiness, then shape a roadmap with a bounded first release. Product strategy, enterprise architecture and integration planning remain part of the work when AI is only one part of the answer.
- Enterprise AI opportunity and readiness assessment
- Product strategy and technology modernisation
- Architecture, integration and delivery planning
Understand the operating reality
Map the customer or employee journey, current systems and decision owners. Surface data gaps, duplicated work and constraints that influence delivery.
A current-state and opportunity mapCompare viable approaches
Assess a product, workflow configuration or custom build. Compare value assumptions, dependencies and operating responsibility, including non-AI alternatives.
An options assessment and target architectureDefine the first commitment
Turn the preferred approach into a bounded scope, delivery stages and acceptance measures. Identify the evidence needed before a wider rollout.
A prioritised roadmap and pilot brief
Business goals → process and product design → data and integration architecture → delivery roadmap. Recommendations connect to implementation artifacts and named responsibilities.
- Process and opportunity assessment
- Build, configure or buy recommendation
- Architecture and dependency map
- Prioritised roadmap and pilot acceptance plan
A roadmap should explain what to build, what to buy and what to stop. AI is evaluated alongside process, software and architecture decisions.
- A discovery recommendation is not a guaranteed business result; assumptions and measures are documented.
- Commercial estimates and architecture choices depend on system access, constraints and the agreed depth of assessment.
Bring to discovery
- Business objective and decision sponsor
- Current workflows, systems and known constraints
- Access to process owners and representative material
Is this only AI consulting?
No. AI leads our focus, while the assessment can cover product strategy, enterprise software, process design, integration and architecture. The recommended answer may include conventional software.
Can you implement the recommendation?
We can scope implementation through an existing product, custom engineering or an AI Squad. Discovery and implementation deliverables are agreed distinctly.
