Skip to content

HOW WE WORK

Five steps, and you can stop after any of them.

A pilot call, an Operations Diagnostic, a pilot, the build in stages, then Run & Improve. The call is free. Each paid step ends on its own, so you decide to go on with what the last step found, not with what a proposal promised.

Book a pilot call

Free · 20 minutes · no commitment

WHAT YOU BUY

Five stages, each one agreed before it starts.

  1. 01 — THE PILOT CALL

    Twenty minutes. No deck.

    You describe a week that went wrong. We ask how the day runs — the orders, the quotes, the paperwork — and say whether an Operations Diagnostic would find anything worth paying to fix. If the answer is no, that is the answer, and the call has still been worth twenty minutes.

    HOW LONG
    About twenty minutes
    WHAT IT COSTS
    Nothing
    WHAT YOU LEAVE WITH
    A straight answer. If it is a yes: what the Diagnostic would map, and what it would cost.
    IF YOU STOP HERE
    Nothing owed, nothing to cancel.

  2. 02 — THE OPERATIONS DIAGNOSTIC

    About two weeks, the whole business mapped.

    We talk to the people who run each part of the business and watch the work as it is really done. We map every system, process and procedure, and how work moves between them. We find the exceptions, where cost and delay live, and baseline one number you already track from your own records.

    We come back with a demo on your own data, a baseline you sign, and a ranked plan with a proposal for a one-workflow pilot. You keep all of it, whether or not you go ahead. Recommending you do nothing is a legitimate outcome.

    HOW LONG
    About two weeks
    WHAT IT COSTS
    Quoted after the pilot call, paid in full, upfront
    WHAT YOU LEAVE WITH
    The map, a signed baseline, a demo on your own data, and a ranked plan.
    IF YOU STOP HERE
    Nothing more. You keep the map, the baseline and the plan.

    The Operations Diagnostic, in full →

  3. 03 — THE PILOT

    One workflow first, run beside the old way.

    We build the first workflow the plan ranked, inside the tools you already own. It starts read-only. It is tested on your real past work, then runs in shadow mode: new and old side by side, until it beats the baseline and you trust it. A person approves every step that matters.

    HOW LONG
    Set in the quote, side-by-side run included
    WHAT IT COSTS
    Quoted on the work, half at the start, half on completion
    WHAT YOU LEAVE WITH
    One workflow working in your own systems, measured against the signed baseline.
    IF YOU STOP HERE
    What was built stays yours, working, in your accounts.

  4. 04 — BUILD IN STAGES

    The next workflow, one stage at a time.

    Each stage is quoted on the work and tested side by side before it goes live, so you never bet the whole operation on one build. Your team is trained in their own workflow, with their own job names, and the playbooks are written for the person doing the job.

    HOW LONG
    Set in each stage's quote
    WHAT IT COSTS
    Quoted on the work, each stage half at the start, half on completion
    WHAT YOU LEAVE WITH
    A system in your own accounts, documented, that your team can run.
    IF YOU STOP HERE
    Every finished stage stays yours, working.

  5. 05 — RUN & IMPROVE

    Watched, so a failure is caught by us and not by your customer.

    Systems drift: a supplier changes a form, a customer changes how they send orders, a new hire does it a different way. Run & Improve is a monthly retainer for the monitoring, the maintenance, and the next improvement on the list, priced against the labour it displaces. It is a twelve-month agreement with sixty days' notice either way, because keeping your systems alive means holding capacity for you. Everything stays yours throughout — every credential, every record, every configuration.

    HOW LONG
    Monthly, for as long as it earns its keep
    WHAT IT COSTS
    A monthly retainer, agreed in advance
    WHAT YOU LEAVE WITH
    One accountable party, and the number reported against your own baseline.
    IF YOU STOP HERE
    Sixty days' notice. Nothing stops working because we are gone.

Twenty minutes, no deck. The call is step one.

Book a pilot call

WHAT WE DO INSIDE YOUR BUSINESS

Seven steps, done where the work is done.

This is forward-deployed work: the way OpenAI and Palantir put AI to work for their largest customers. The person who builds it works inside your business.

  1. 01

    Get inside the work

    On site. We talk to every process owner and watch the job as it is really done.

  2. 02

    Map every system

    Each piece of software, spreadsheet, inbox, portal and paper form, and how work moves between them.

  3. 03

    Find the exceptions

    What happens when things go wrong. That is where the cost and the delay live.

  4. 04

    Agree one number

    A baseline from your own records, signed before anything is built.

  5. 05

    Build around the core

    Your ERP and core systems stay. Small tools are improved, or replaced when something better fits.

  6. 06

    Test, then run side by side

    Tested on your real past work, then run beside the old way until it beats the baseline.

  7. 07

    Stay and run it

    We watch it every month and keep improving it, so it keeps working long after go-live.

Forward-deployed work, explained →

Inside your operation. In your systems. Measured.

Inside the work

The person who builds it comes to your floor and watches the job first, wherever the work is done.

  • We sit with the people who do the work before we build anything.
  • Start with one job, done properly.
  • A person approves every step that matters.

In the systems you already run

We connect to the software and tools you already have. We do not ask you to change them.

  • Your systems stay. Nothing is replaced.
  • Read-only first. Nothing changes until you approve.
  • Everything we build stays in your accounts.

Measured against your own records

We measure it before we build it, with a number you already keep.

  • One number, agreed before the build.
  • Tested on your real work before it goes live.
  • New and old run side by side until you trust it. Then we stay to run it.

The controls, at every step.

One number, agreed in advance

Every step names the number it is trying to move before it starts, taken from your own records, so success and failure are legible to both sides.

Read-only first

We start with read-only access to your ERP and inboxes. Nothing is written into your systems until you approve it.

Side by side until you trust it

Before go-live it runs in shadow mode on your real work: the new way and the old way together, compared, until you are ready to switch.

A person approves every step that matters

An order entered, a price sent, a date promised: each waits for a person. Your customers are told when they are dealing with AI.

Everything stays in your accounts

The configurations, workflows and documentation live in your accounts, with access scoped to the work and yours to revoke. Never banking, payroll, or stored customer payment details.

We stay to run it

If we build it, we watch it. A system that fails silently is worse than the manual process it took over.

Bring one bad week.

Nothing needs preparing. The call goes faster with these to hand.

  • One bad week — the order keyed wrong, the shipment that could not be found, the invoice that went out wrong.
  • Any numbers you already track: orders a week, how many arrive by email or PDF, hours the desk spends keying them. Rough is fine.
  • Who runs the desk, and what they do when it gets busy.
  • The ERP and the tools you already pay for. Most of what we fix is connecting what is there.

How each step is priced →

The questions people ask before they book →

Bring us one bad week.

Book a pilot call. Tell us where the week goes wrong, and we’ll tell you what an Operations Diagnostic would map and measure.

The first paid step has a clear scope and an end date, and the pilot after it is one workflow — not a platform, and not an open-ended retainer.

Book a pilot call

Free · 20 minutes · no commitment