What we build
One recommendation, not a menu.
After a discovery call we propose one scope, with the reasoning behind it. A menu transfers the decision to the person with less information — which in this arrangement is you, and that is not a fair trade.
Examples of what the same engine gets configured to do.
These are the processes we get asked about most. They are not fixed packages, and the list is not a claim that we have built every one of them for a customer — see where we actually are.
-
Lead and enquiry handling
Capture, verify, qualify, route and follow up on every enquiry — including the ones that arrive at 11pm.
-
Customer intake
One intake that asks once and files it correctly, instead of the same details being collected three times and typed in four.
-
Scheduling
Qualify first, then move the right people toward an appointment, so the calendar fills with work you actually want.
-
CRM and record automation
Records created, updated, tagged and routed automatically, so the file exists before anyone has read the email.
-
Follow-up
Capped, with stop conditions, and never a machine that nags forever. Nothing goes cold because somebody meant to call back.
-
Document and data handling
Pull information out of documents and move it into the systems where it belongs, rather than re-keying it by hand.
-
Internal workflow
Connect the tools your team already runs so the same information stops being entered in four places.
-
Screening and duplicate detection
Check an arriving enquiry against the records you already hold, and report what was found — by name, never as a cleared status.
Seven stages, and the methodology is as much of the product as the code.
-
Discovery — 45 minutes
How the business actually operates. What arrives, who touches it, where it stalls, what it costs when it does. We finish by doing the arithmetic on the gap using your numbers, not ours.
-
Design
Map the workflow and decide what should be automated — and explicitly what should not. Includes the failure design: what happens when each step breaks and who finds out, written down before anything is built.
-
Build
Develop and configure. Reusable components where they exist, work specific to your business where they do not.
-
Integrate
Connect the tools you already run. This is where engagements actually get blocked, so we check feasibility before quoting rather than after.
-
Test
Simulations, edge cases and failure scenarios. Every misclassification we find becomes a pinned regression test in the same change, so it cannot come back quietly in three months.
-
Deploy — supervised
Live, with you watching it. Anything misclassified gets corrected the same week it appears.
-
Optimise
Monitoring, and improvement as the business changes. This is the part the monthly pays for.
What the engagement assumes from you
Stated here so it never becomes a surprise halfway through a build:
- Software already in place that the system can write into, and an administrator who can authorise an API connection
- A decision-maker available for a 90-minute configuration session, in person where that is practical — the decisions cluster, and ninety seconds in a room beats three weeks in an email thread
- Your rules and thresholds written down, which usually means we write them down together in that session
- Approval of every message that goes out under your name, read aloud before go-live
- Two weeks of a human reviewing output before the system runs unsupervised
An implementation fee plus a monthly. Never a one-time build.
The implementation fee buys discovery, the workflow design including its failure behaviour, the build, the integration, testing, and a supervised launch.
The monthly buys monitoring, fixes when a vendor changes something underneath us, retuning as your business shifts, and the fact that somebody is watching. An unmaintained automation is a liability handed to a client — a workflow that quietly deactivates six months after handover, with nobody looking, is worse than never having built it. So we do not sell one and walk away, at any price.
The monthly is not paying for compute. Infrastructure runs a few tens of dollars a month and the variable cost is three to five cents per enquiry — measured, not estimated. We will tell you the real figure on the call rather than let you assume AI carries unpredictable per-use billing, because a prospect who suspects they are paying a margin on API calls negotiates against the wrong thing.
Why there are no numbers on this page
There is validated pricing in exactly one vertical, worked out against delivered hours. Outside it we would be guessing, and a figure published today that moves in a proposal next week is worth less than no figure at all. You get the number, with the scope it covers, before you commit to anything.
What we will say plainly: price this against the alternatives, not against our cost. A part-time hire costs more than the monthly and does not work at 11pm. An answering service runs monthly, forever, and does none of the above. And if a packaged product genuinely fits your process, buy it — we are for the case where it does not.
Terms: twelve-month initial term, monthly thereafter. Additional workflows are scoped as separate implementations rather than absorbed.
Start with a 15-minute call.
Fifteen minutes on the phone to work out whether there is anything here worth building — and if there is not, we will tell you that rather than sell you something. The arithmetic comes in discovery, above.