Skip to content

AI automation

If you can describe it, we can probably build it.

An AI receptionist that answers at 2am. A booking agent that fills next week. Intake that writes itself and invoices that chase themselves.

Built for law firms, and it runs a dental practice or a thrift store just as well.

How we think about it

This is plumbing, not magic.

Every automation here is a small, dull machine that does one thing on time.

The cost of a task isn’t the hours

An hour a week is fifty hours a year, and that’s the smaller half. The real cost is the interruption.

We build three things, not thirty

Most of what you do by hand should stay that way. We look for work that repeats.

Bought tools break where you’re unusual

Off-the-shelf AI works until it meets the exception that makes your firm your firm. We build inside the tools you already pay for.

Ethernet cables patched into the numbered ports of a network panel.
A patch panel does nothing clever. It gets what arrives where it belongs, every time.

The catalogue

Everything we have built before.

Twenty-nine builds, grouped by the part of the week they take off your hands.

A glass office tower at night with most of its floors still lit.
Somewhere in a building like this, a phone is ringing at 2am.

How a build runs

Five steps, and you can stop after two.

01

Discovery

An hour mapping your week: what happens, who does it, how often it goes wrong.

02

The shortlist

The three tasks worth automating, what each saves and what each costs.

03

Build

In your accounts, on your data, at a fixed price. You watch it run on real cases first.

04

Pilot with a human in the loop

It drafts rather than sends. Nothing goes autonomous until you’ve watched it be right.

05

Handover or hand-holding

Documented and handed over, or kept running by us monthly. Change your mind later.

A desk phone with its handset in the cradle on an empty office desk.
What changes on the desk is small: same number, same greeting, answered every time.

Guardrails

The part attorneys read twice.

An automation that’s quietly wrong is worse than none, so these four get set first.

A human approves anything that matters

Default is draft-and-wait. Anything that sends, books or spends goes autonomous only after you’ve watched it behave, and one click stops it.

It runs in your accounts

Your CRM, your number, your calendar, your card. If we part ways, the automation and its output stay with you.

Grounded, not guessing

Anything client-facing answers from your own documents. We’d rather it say "let me get a human" than invent something confident and wrong.

The confidentiality conversation happens first

For regulated practices we walk through what data touches which system before anything switches on.

Pricing

Two numbers, and an optional third.

Priced as a project, not a retainer. Nobody should pay an hourly meter to find out whether a build works.

Discovery is paid, and it stands alone

A fixed fee for the mapping hour and the written shortlist that follows.

Scope first, then a price

We work out what it has to do and what it has to talk to, and only then quote it — as one figure rather than by the hour. A receptionist for a two-person firm and one for a busy clinic are not the same job and should not carry the same number.

Monthly is optional

Maintain it with us, or take the documentation and run it in-house. Platform costs sit on your card at cost.

Questions

The things people ask before they say yes.

No, and be suspicious of anyone who promises it will. It takes the repetitive work off your team so the same people handle more.

Yes. We tell them, and it offers a human immediately. Pretending otherwise annoys the exact person you were trying to win.

It will, occasionally, so nothing important runs unsupervised. Everything is logged, and it escalates to a person rather than guessing.

No. Building one is usually the first automation, because everything else needs somewhere to put its results.

Yes. An AI receptionist doesn’t care what you sell. Contractors, clinics, salons and retail run the same plumbing.

Tell us the task that eats your week.

Describe the thing somebody in your office quietly resents. We’ll tell you whether it’s worth automating.

You keep the shortlist whether or not we build anything.

See the workBook a call