The Disciplines Plate I AI Adoption & Implementation

AI Adoption &
Implementation.

AI that actually does a good job for your business.

Mostly document search, customer support and first drafts, for small businesses across Suffolk and Cambridgeshire. Which model I use, I decide by testing several of them against the real task.

Return to the workshop

AI can save a small business hours every week, and it can do a few things which were not possible at all a couple of years ago, but most of the skill is in choosing which problems to point it at and in recognising the ones that are better solved the old-fashioned way.

So before I build anything I look at which jobs are repetitive enough, and tolerant enough of the occasional wrong answer, to be worth handing over. Sometimes none of them are, and then I will tell you so, because I am not going to put an "AI-powered" sticker on a tool that worked perfectly well without one.

§ The work

  1. I.

    Intelligent document search

    You ask a question in plain English and you get back the passage which answers it, from across your emails, PDFs and shared drives, including the documents nobody remembers writing. The technique behind it is called retrieval-augmented generation, or RAG.

  2. II.

    AI customer support

    A chat or email assistant which answers from your own material, so it takes the common questions at any hour, sorts what comes in and passes anything that needs a judgement call on to a person. The aim is fewer tickets, but without bad answers going out under your name.

  3. III.

    Data analysis & insights

    Getting something useful out of the spreadsheets nobody opens, for example which customers are buying less or what changed in the last quarter, without having to hire a data-science team or pay for a six-figure BI platform.

  4. IV.

    Content generation & summarisation

    First drafts of the repetitive writing, such as product descriptions, meeting notes or research digests, and a one-page summary of the 40-page report which nobody was going to read. What goes out is still decided by a person.

§ The process

  1. Step 1

    Discovery

    I start by looking at where your time goes: which jobs are done by hand every week, and which information is hard to find at the moment you need it.

  2. Step 2

    Opportunity analysis

    I pick out the few uses of AI which make sense for your business and your budget. Some of them need building and others are just an off-the-shelf tool set up properly, and I will tell you which is which.

  3. Step 3

    Proof of concept

    Before anything big gets built I make a working prototype, which you try on your own work, and then you decide whether to go further before you have spent much on it.

  4. Step 4

    Implementation

    I build the full version and connect it to the systems you already use, then show your team how it works.

  5. Step 5

    Refinement

    I keep an eye on how it performs and I adjust it as your team finds the rough edges, and the ongoing support is included in the price.

§

Questions, answered

Before you
have to ask

How do I know whether AI is worth it for my business?

Usually by looking at where the same question gets answered fifteen times a week, or where somebody reads through a document only to find one number, because those are the jobs AI is good at. If I look and the answer is that AI won't help, then I will say so.

Will my data be used to train a model?

No. Where I use a hosted model, it is one whose terms exclude training on customer data, and that is in writing. If that is not acceptable for you, the work runs on a model hosted on infrastructure you control.

What will it cost?

The prices on this page are starting points, so that you can tell in a minute whether you are in the right range. I then quote each job once we have talked it through, because the price depends on the shape of the job, and at this point nobody knows that yet. The first half hour is free, and usually it settles whether this is a small piece of work or a large one before any money is discussed.

Who owns the code when it is finished?

You do, outright, including the source. I hand it over with the runbooks you need to keep it running, and there is no licence to renew and no seat to pay for. When I say commissions, not subscriptions, I mean it literally.

What happens if you are unavailable?

I build everything so that somebody else could pick it up: common technologies, readable code, written runbooks and nothing proprietary. It is a fair question to ask of a one-person business, and the answer lies in how the work is built.

Do you work outside Suffolk and Cambridgeshire?

Yes. With local work you get the option of having me in the room, which is worth more than it sounds for the first conversation and for the handover, and for most clients everything after that is done remotely anyway.