Practical AI advisory

Find the AI workflow worth testing first.

Review up to three recurring workflows, compare the opportunities and see one focused demonstration before deciding what to build.

Discuss an AI workflow

Up to 3 workflows1 focused demonstrationA measured pilot plan

The situation

When this is useful

Your team is experimenting with AI but has no shared priorities.
Reporting, document preparation or knowledge retrieval takes repeated manual effort.
You need to compare potential value with data readiness, review effort and operating cost.
A promising demo needs a practical route to a measured pilot.

The output

What you receive

Workflow shortlist — up to three workflows mapped around users, inputs, handoffs and outputs.
Opportunity assessment — value, feasibility, data needs and material uncertainties.
One focused demonstration — a bounded example using synthetic data or material approved for the agreed tools.
Pilot brief — scope, owner, success measures, review points and next steps.
Findings discussion — a walkthrough of the recommendation and the choices it depends on.

A focused engagement

Small enough to test.
Clear enough to evaluate.

Format: Remote, fixed scope. Up to three workflows and one focused demonstration.
Delivery: Timing is agreed after the initial scope review.
Fee: Quoted for the agreed scope before work begins.

Scope and boundaries

Production integrations, software rollout and ongoing operation are scoped separately. The review establishes a practical pilot proposal; deployment decisions remain with your organisation.

A concrete example

A weekly management update.

A team combines status notes, delivery updates and open risks into a weekly report. The review examines which inputs are reliable, where drafting can be assisted and what the owner still needs to verify.

Illustrative pilot measures

Time spent preparing and checking the update.
Number of unsupported or incorrectly attributed statements.
Rework required before the report is useful.
Whether the owner chooses to keep using the workflow.

Illustrative example — fictional scenario

AI pilot brief

Workflow: Weekly management update.
Inputs: Approved project notes and an agreed reporting format.
Assistance: Draft a summary with links to source statements.
Owner check: Verify status, dates, risks and unsupported claims.
Pilot measure: Preparation time, review effort and corrections required.
Next decision: Continue, adjust the workflow or stop the pilot.

Before we start

Practical questions.

Do we need to choose an AI platform first?

No. The review starts with the workflow and its constraints. Existing approved tools are considered before any new purchase.

Do we need to share confidential documents?

Initial scoping can use a description of the process. Any sample data and tools used for a demonstration are agreed before the work starts.

Can this include an internal knowledge assistant?

Yes, as a bounded use case. The review looks at source quality, access needs, answer checking and the work required to keep the knowledge current.

Can this include agents or MCP?

Yes, where tool access is relevant to the workflow. Ownership, permitted actions and production requirements form part of the assessment.

Will you build the production system?

Implementation needs a separate scope, technical delivery plan and operating owner. The review gives you a basis for that decision.

A practical next step

Which workflow is worth testing?

Tell me what you are working through, what is at stake and when you need to act.