How it works

Two weeks. One task. People in charge throughout.

Short on purpose: long enough to measure a real change, short enough that it's an easy yes.

The fortnight

From first call to a decision you can defend.

  1. Before we start

    A 30-minute call

    You describe one task your team repeats. We check it is a good candidate (it happens often, it follows a recognisable pattern, and someone looks after it) and agree who we will speak to. If it is not a good candidate, we say so and stop there.

  2. Days 1–3

    See the task as it really happens

    We sit with the people who do it and time the real thing, not a demo. Every step, hand-off, wait and exception goes on one page that the team agrees is accurate.

  3. Day 4

    Decide the smallest fix

    Standardise, digitise, automate the rules, add an AI draft with human review, or leave it alone. You see the reasoning for each option and choose. Nothing is built without your sign-off.

  4. Days 5–12

    Make one change and run it

    We build the change in tools you already use where possible. Your team does the task as normal, and we time it the same number of times as before.

  5. Days 13–14

    Keep it, adjust it, or undo it

    Before-and-after figures from your own work, a decision record in plain English, and a handover so your team can run it without us. If the change did not help, we undo it.

Two colleagues passing documents across an office desk
What you get

Small, checkable outputs.

  • A one-page map of the task, agreed with the people who do the work.
  • A decision record: what changed, why, who approves outputs, and what would make you undo it.
  • Before-and-after timing from your own work.
  • A handover note, so your team can run and adjust the change without us.
Ground rules

What we will and will not do.

We will
  • Tell you when a task isn't worth automating.
  • Keep a named person accountable for every output.
  • Use the smallest change that works, and make it reversible.
  • Document everything so you are never dependent on us.
We will not
  • Let AI send, pay or decide anything on its own.
  • Put personal data into AI tools without a data-protection check.
  • Quote savings before we have measured them.
  • Lock you into a platform or a contract you cannot leave.
Questions

What people usually ask.

Do we need to buy new software?

Usually not. Most first improvements use tools you already pay for: forms, spreadsheets, your email or document system. If something new genuinely helps, we explain why, what it costs and how to leave it.

Will you use AI on our data?

Only where it earns its place, and never on personal or sensitive data until a data-protection check confirms what is allowed. AI drafts; a named person reviews and decides. Many tasks need no AI at all.

What do you need from our team?

One person who looks after the task, two or three people who do it, and about two hours of their time across the fortnight, mostly during mapping and the final review.

What does it cost?

We agree pilot terms case by case and confirm them in writing before any work starts. Nothing is charged without a written agreement.

What if the change does not work?

Then the pilot has done its job: you know for sure, and we undo the change. You keep the map of the task and the record of decisions either way.

Bring one task. Leave with a decision.

A short call about the work your team repeats most. If automation isn't the answer, we'll say so.

Talk to usTry the time-back check