The process
From first call to a live system, usually in one to three weeks.
Most AI projects fail in the same place: nobody agreed what the system was allowed to do on its own. So the process below spends its effort on that question early (what it answers, what it books, and exactly where a person takes over) and treats the building as the straightforward part, because by then it is.
This is how a build actually runs, including the parts that are on you rather than on us.
The call
Free, 30 minutes, no pitch.
We ask how enquiries actually reach you today (the phone, WhatsApp, a web form, Instagram DMs, a portal you log into) and what happens to each one after it arrives. Most of the useful detail comes out of the boring questions: who sees the message first, what they do next, and what they do when they are busy.
By the end of the call we should be able to say which part of that is worth automating and which is not. Sometimes the honest answer is that the fix is a process change rather than a system, and we will say so.
The build
One to three weeks, with progress you can see every week.
We scope the system as a set of concrete behaviours: what it answers, what it collects, what it writes into your calendar or CRM, and exactly where it stops and hands over to a person. Those handover points are agreed before anything is built, because they are the part that determines whether you can trust it.
Then we build it on the platforms your business already uses rather than moving you onto ours. What moves the timeline is almost never our build time; it is access. If the system has to read and write a booking platform or CRM, we need working credentials and a clear answer on the rules early, or testing stalls.
Going live
Tested against real messages before it answers a real customer.
Before launch the system is run against the kind of messages you actually receive: the vague ones, the ones with three questions in a single sentence, the ones sent at 1 AM. We check what it does at its limits, not just on the clean path.
It goes live on your own number and your own accounts. You keep the customer relationship, the message history, and the data; nothing sits behind a login only we hold.
After launch
Monitoring, adjustments, and support as the business changes.
Real conversations always surface things a scoping call cannot: a question nobody predicted, a phrasing the system reads wrong, a step that turns out to matter more than it looked. We watch the first weeks closely and adjust.
After that the system needs occasional attention rather than constant work: when your services change, when your hours change, when a platform changes its rules. That is what ongoing support covers.
What we need from you
Four things, and the build moves fast.
Access, early
Credentials for whatever the system has to read or write: the booking platform, the CRM, the WhatsApp number. This is the single most common cause of a slow build.
Your real answers
What you actually tell people about pricing, availability, preparation and location. The system answers from your information, so it can only be as accurate as what you give it.
One decision-maker
Someone who can settle a question the same day. Where a build stalls, it is usually waiting on an answer, not on code.
Where a human takes over
The cases you want a person to handle: anything urgent, anything clinical, anything where a wrong answer costs you. We build to that line.
Before the call, these help
Start with the call.
Thirty minutes, no pitch. Tell us how enquiries reach you today and we will tell you what is worth automating, including when the answer is that nothing is.