Custom automation
Every business runs on a few tasks that only happen when somebody remembers.
They get done in a quiet week and skipped in a busy one, which is exactly backwards, because the busy week is when they were worth the most. Those are the ones worth building, and which ones they are is not obvious from inside.
The expensive mistake is automating the wrong thing, competently
Most owners can name the task that annoys them most within about four seconds. It is rarely the one costing the most, and the gap between the two is where automation projects go to be disappointing. Something gets built, it works, everybody agrees it is clever, and nothing measurable changes.
The irritating task is loud. The expensive one is usually silent.
A leak that costs real money tends to make no noise at all, because the thing it loses is a customer who simply went elsewhere and never told you why. Nobody complains about the follow-up that was never sent. So we measure from the outside before building anything, which is what the audit is.
What this work usually turns out to be
Four shapes cover most of it. They look different from business to business and they are the same four underneath.
The thing that only happens when somebody remembers
The quote that gets followed up on day three, the review request after a job closes, the check-in with a customer who went quiet. Every business has three or four of these and every one of them happens perfectly when things are calm and not at all in the week you were busiest, which is the week it was worth the most.
The same fact typed into three places
A new customer entered in the scheduling system, again in the accounting software, again on a whiteboard. Each copy is a chance for the three to disagree, and once they disagree nobody can tell which one is right without asking the customer. This is fixed by making one of them the record and having the others read from it.
The handoff that goes through a person for no reason
A form submission somebody reads and re-enters. A booking that arrives by email and gets copied into a calendar. Work that arrives in one system and has to be carried by hand into the next one, where the person carrying it adds nothing except the risk of dropping it.
The report that gets built the night before it is needed
Numbers pulled out of two systems and reconciled in a spreadsheet, monthly, by whoever is least able to refuse. Usually the answer is not a dashboard. It is deciding which number actually gets acted on and putting that one where it is already being looked at.
Work finished, not work described.
Most tools that touch this space stop at the moment of describing. Something happens, a record of it appears, and every piece of work that event created is still sitting there waiting for a person. A transcript is not a booking. A notification is not a follow-up. A summary is not a job moved to the next stage.
What we build hooks into the systems you already run and finishes the work.
That is only possible because it is built from what your business knows and has never written down: which jobs you take, what you charge to come out, who gets called when something is urgent, what has to be true before an invoice goes. Nothing bought off a shelf has any of that, and no amount of configuring gives it to them.
How it runs, and why the order matters
The audit names the work. Five touchpoints checked the way a customer meets them, and a written report ranked by what each gap costs. It ends with the leak worth closing first, which is the input to everything below. $1,000, and it is the only fixed price in this sequence.
The quote follows the audit. Once there is a named thing to build and we have seen the systems it has to touch, the number is a real number rather than a range hedged against unknowns.
We build the smallest useful piece first. One leak sealed, running in your business, before anybody commits to the next one. This is slower to describe and faster to be worth something, and it means you can stop after the first piece if it did not do what we said.
Then the next one, in cost order. Contain, then expand. A business that grows while it leaks just leaks faster.
No price on this page, on purpose
Everything else we sell has a number on its page. This one cannot, because the work is defined by what the audit finds, and the range from one connection between two systems to rebuilding how jobs move through a business is too wide for a floor to mean anything.
The audit names the work. The quote follows it. That is the whole pricing policy and it is not a way of avoiding the question.
What we will do on a first call is tell you roughly what size of thing you are describing, and tell you when what you are describing is smaller than you think.
When not to buy this
If your problem is a known one with a page and a price. Unanswered calls, texts being dropped by carriers, assistants naming your competitors: those are diagnosed already and you should buy the specific thing rather than a scoped engagement.
If the process you want automated is about to change. Building around a way of working you are already unhappy with buys you a faster version of the thing you were going to replace. Settle it first.
If nobody on your side can answer questions for a few hours. This work runs on things only your team knows, and an engagement where those questions go unanswered for a fortnight becomes us guessing at your business in an expensive way.
Questions
What does this cost?
It is quoted after the audit, and there is no starting-from number on this page on purpose. The work ranges from a single connection between two systems to rebuilding how jobs move through a business, and a floor quoted here would be a number you had to be talked out of before we could talk about what you actually need.
Why does it have to start with the audit?
Because the expensive mistake in this work is automating the wrong thing, competently. The task that annoys you most is often not the one costing you most, and the two are surprisingly hard to tell apart from inside. The audit measures it from the customer's side and names what is worth building before anybody quotes a build.
Do I have to change the software I use?
Almost never, and we push back when somebody suggests it. The systems you already run hold your history and your team already knows them. This work connects them and adds what is missing between them. Replacing a working system to make an automation easier is charging you for our convenience.
What happens when it breaks?
It gets fixed, and how that is covered is part of the quote rather than an afterthought. Anything that runs unattended eventually meets something it was not built for, and a system nobody owns after launch is a system that quietly stops one Tuesday and gets noticed a month later.
Is this AI, or is it scripts?
Whichever the job needs, and most jobs need both. A model is the right tool for reading something unstructured, deciding what a message is about, or writing a reply. It is the wrong tool for moving a record between two systems reliably, which is ordinary engineering. Anyone selling you a model for the second kind of work is selling the fashionable answer rather than the correct one.
How is this different from the phone agent or the chat?
Those are two specific automations we have already built and can price on a page. This is the same discipline pointed at whatever is leaking in your business rather than at a known problem. Plenty of engagements start as one of those and become this once the audit shows where the work is actually piling up.
What gets done only when somebody remembers?
Answer that and you have named a candidate. The audit tells you whether it is the one worth building, or whether the money is leaking somewhere quieter.
Scoped after the audit. See everything we do if you would rather buy something already defined.