Proving Ground
How we work

Four phases, and a real decision at the end

Understand the business, commit to one workflow, prove it small, decide on evidence. What changes between clients is which workflow — not the shape.

  1. Week 1

    Assessment

    Work out what is actually true about how your business runs today.

    • Walk the workflows with the people who do the work
    • Check what data exists and whether it can be got at
    • Name the constraints early — privacy, contracts, appetite
  2. Week 2

    Strategy

    Sort the candidates honestly, then commit to exactly one.

    • A ranked shortlist with the effort and risk beside each
    • One workflow chosen to start
    • The baseline measured before anything is built
  3. Under 8 weeks

    Pilot

    Short build cycles, with a person in the loop from day one.

    • A working version in front of real users early
    • A review gate before anything reaches a customer
    • Paid training time for the people whose work changes
  4. The end

    Decide

    A decision made on evidence, including the option to stop.

    • Measured result against baseline, written down
    • Scale, adjust, or stop — a real option, not a formality
    • If you scale, we stay on to keep it honest
Oversight

A person stays in the loop

Autonomy is a dial, not a switch. Every workflow starts fully reviewed and only loosens where the measured error rate earns it.

  1. Draft, don’t send

    The system prepares the reply. A person approves it. That one rule removes most of the ways AI embarrasses a business publicly.

  2. Log every decision

    What the system saw, what it produced, who approved it — so a disagreement later is settled by looking, not arguing.

  3. Loosen on evidence

    Once a step’s error rate is measured and low, it can run unattended and be sampled instead. Your call, with numbers in front of you.

  4. A stop switch that works

    Your team can turn any workflow off without calling us. If a system can only be stopped by its consultant, it owns you.

Why we weight this heavily

70%

of what determines an AI transformation is people and process. Algorithms account for 10%, technology and data for 20%.

Boston Consulting Group, The 10-20-70 framework, drawn from roughly 1,500 organisations implementing AI (2024)

more likely to meet their objectives: initiatives with excellent change management, versus those with poor change management.

Prosci, Best Practices in Change Management, 12th edition (2024)
Privacy

Your data, kept to the minimum, in accounts you own

Collect less

A workflow gets only the fields it needs, held only as long as it needs them.

You own the accounts

Built under your billing where possible. If we part ways, it keeps working and you keep the keys.

Named services

A plain list of every third-party service involved and what each one sees.

General practice, not legal advice. For a regulated workflow we'll tell you when you need a lawyer rather than a consultant.

Start here

Bring one bottleneck. Leave with a straight answer.

A 20-minute call, no deck. We'll say whether AI is the right tool for it — including when it isn't.

Not ready to talk? Take the scorecard first — no email needed to see your score.

No newsletter, no sequence. Your note goes straight to the team.