AI that fits the business you already run

Most of what gets written about AI is aimed at companies with an IT department and a budget line for experiments. If you are running a twelve-person business, the useful question is narrower: which parts of the week are being eaten by typing things up, looking things up, and writing the same email again?

That is where this starts. Not with software, and not with a demo. The goal we aim at is simple enough to measure: get an hour or two a day back per person, using the systems you already pay for.

What it actually looks like

We work mostly with trades and service businesses (HVAC, electrical, mechanical, construction) because that is where the paperwork-to-revenue ratio is worst and the fix is most obvious.

Before the visit. A summary of everything on file for that customer, ready on the phone, instead of a technician scrolling through job history in the van.

After the visit. The technician records a thirty-second voice note, “replaced the capacitor, unit’s running, recommended a service plan”, and the service report, the customer email and the invoice notes are drafted from it. The paperwork that used to happen at nine at night happens on the drive back.

Quotes and proposals. A few bullet points become a first draft of an estimate in your own format and language, ready for you to price and correct.

The follow-ups that never get sent. Drafted automatically, so the quote from three weeks ago actually gets chased.

Dispatch and scheduling notes summarised instead of read in full.

Most of this needs no new hardware and no new system. What it needs is your existing information to be tidy, your permissions to be right, and the connection between the two to be secure.

Which is where the security part matters

An AI assistant is only useful if it can see your business information. That is exactly what makes it a security question: what it can see, who is allowed to ask it, and where the answers go.

Get the permissions wrong and you have built a very efficient way for the wrong person to read the payroll file. This is the part most AI projects skip, and it is the part we are qualified for. It is the same CASP+ background behind the security work we do for our managed clients.

How we start

1. AI readiness assessment. What systems you run, where the information lives, who can see what, and which tasks are actually worth automating. Ends with a written roadmap and an honest answer, including “not yet” if that is the answer.

2. A first project, one department. Pick the area with the most repetitive work. Tidy the permissions, organise the data, configure the tools, train the people who will use it, and measure whether the time saving is real.

3. Ongoing, if it is worth it. New automations as you find them, reviews as things change, and training for new staff.

You can stop after step one. Plenty should.

We are our own first customer

We do not have AI client case studies yet, and we would rather say so than invent one.

What we do have is our own business running on this. The company knowledge base, the accounting and calendar systems connected so they can be queried directly, the marketing drafted and published on a schedule without anyone opening a browser, and this website, which was migrated, rebuilt and tested largely by AI working under supervision.

That is the honest position: we have done this to ourselves first, at our own expense, and we can show you the working and the parts that did not work. When we have client results to publish, they will appear here with the client’s name on them.