AIgment

AI should shadow your work before it touches it.

For teams buried in documents, inboxes or data entry, or holding a list of AI ideas and no way to choose. We watch how the work runs today, test AI on your real cases and build what passes. You can stop after any stage.

You see the scope and the price in writing before any paid work starts.

Four stages. You can stop after any of them.

Each stage answers one question and ends with something you keep. Start wherever fits. The room below follows a made-up example, a wholesaler whose staff type supplier invoices into the ordering system by hand.

1 Assessment

What is worth testing, and what should wait?

Fee
Fixed, agreed before we start
Time
2 to 3 weeks

We sit with the people who run the workflow and answer four questions. Which ideas are worth it? Can your data support them? Where could they go wrong? Who will run them? If a plain script beats AI, we'll say so.

You get a ranked idea list, a value estimate with its assumptions, a risk list and a 90-day plan.

You can stop here. Your own team or another partner can run with the plan.

2 Pilot

Does it work on your own cases?

Fee
Fixed, agreed before we start
Time
3 to 5 weeks

With you, we pick real cases and agree the pass mark before we write any code. Then we build a small version and test it on those cases, failures included.

You get a technical design, a working prototype and a results report on your own cases.

You can stop here if the results don't justify a build. You've spent a pilot's budget, not a project's.

3 Implementation and handover

How do we put it into daily use?

Fee
Fixed, agreed before we start
Time
1 to 3 months

We build it into your systems, with the tests, human checks and limits agreed with you. The code lives in your repository and runs in your own cloud account.

You get working software, tests, a runbook and a handover session. The pilot's test cases stay with you, so your team can re-run them after any change.

You can stop here. Your team runs it from now on.

4 Ongoing development

Who looks after it once it's live?

Fee
Fixed monthly, or hours as worked
Term
Month to month

Suppliers change layouts and models get replaced. We fix what breaks, retest, and add what you ask for. Or we work alongside your developers.

You get a short monthly note, plus a fixed monthly scope or authorised hours invoiced as worked.

You can stop any time with 30 days' notice. Everything built stays with your team.

Same story, on paper.

The example is a wholesaler. Three people spend part of every week typing supplier invoices into the ordering system. The owner has a list of AI ideas and no time to test any of them.

An illustration, not a client. The same method fits support inboxes, contracts, reports and sales data.

Assessment, supplier ordering

4 ideas, ranked
  • Read supplier invoices into the ordering systemPilot thisAbout 400 a week, and every one can be checked against its PDF.
  • Draft replies to routine customer emailsLaterUseful, if a person approves each reply before it goes out.
  • A chatbot on the websiteWaitNobody owns it yet, and nobody has said what success looks like.
  • Predict which quotes will closeNot nowThe sales history has too many gaps to learn from.

Top idea, rough value. Three people each spend about 6 hours a week on invoices. If a pilot reads 9 in 10 correctly, that frees roughly 15 hours a week. Assumes volume stays near 400 a week.

Pilot results

30 real invoices, 12 suppliers

Pass mark, agreed before the build. At least 27 of 30 read correctly, and anything it can't read is flagged for a person, never guessed.

28Read correctly

2Flagged for a person

0Wrong and not flagged

The two flagged were a faded scan and a delivery note with handwritten changes.

Passed. 28 correct against a pass mark of 27, and nothing wrong went through unflagged.

In the example the owner decides to build it. Once it is live, the reader does the typing and the three people keep the judgement. Anything it is unsure of waits for one of them.

What you can hold us to.

These apply to every engagement, large or small.

AIgment

  1. 1

    The price is agreed before any work starts.

    You get a written scope and fee before paid work begins. Anything outside the scope is quoted separately before we do it.

  2. 2

    Every estimate shows its assumptions.

    Cost and value estimates list their assumptions and how uncertain they are. They help you decide. They are not a promise of savings.

  3. 3

    We tell you when we'd profit from our advice.

    If we recommend work we could sell you, we say so. We disclose any referral fee or resale arrangement.

  4. 4

    You own what you paid for.

    The custom work you paid for is yours. If it uses tools we already own, you get a licence to run and maintain it, and to pass it on to your own customers where we agree that.

  5. 5

    Your data stays under a written agreement.

    Before we work with real data, we sign an NDA and a data processing agreement with you that names every provider involved. We use provider terms that exclude training on your data.

  6. 6

    Support only when you buy it.

    Hosting, monitoring and support are separate. If you buy them, the response times are written down. If you don't, we're not on call.

Before you email.

The questions we would ask, too.

Do we need to know what to build?

No. A list of ideas is enough, and so is one process everybody complains about. The assessment sorts out what's worth doing.

Does an assessment commit us to more work?

No. It is a complete piece of work. Use the plan with your own team, another partner or us.

Who will we work with?

One engineer, from the first email to the handover. The person who scopes the work is the person who does it.

What happens to our data?

Before any real data moves, we sign an NDA and a data processing agreement with you that names every AI provider involved. We use provider terms that exclude training on your data, prefer processing in the EU, and can work entirely inside your own cloud account. When the work ends, we delete our copies and confirm it in writing.

Can you work with the tools we already have?

Usually. We start from the systems you already run, and what they let us access decides what's possible.

What if the pilot doesn't work?

Then you know before paying for a full build. We agree the pass mark with you before the pilot starts, so there's no arguing about the result afterwards.

Can you work inside our team?

Yes. We can work with your developers on hours agreed each month, invoiced as worked or reserved in advance.

Is there a smaller way to start?

Yes. A written second opinion on an AI feature or automation your team already built. It says what will break and how to fix it, and takes about a week.

Tell us about one workflow.

A few sentences is enough. We'll tell you whether an assessment is worth doing, or whether you need something else.

  1. You email. The letter here is a starting point.
  2. We reply within 2 working days and suggest a 30-minute call.
  3. Within 3 working days of the call, you get a written scope and a fixed price. If you accept it, work usually starts within 2 weeks.
To
hello@aigment.dev
Subject
A workflow we want to improve

Hello,

Today, .

It's done by , about .

We've already tried .

Open this as an email

Fill in the blanks, or ignore them and just write.