← All buildsThe App You Want
Build story

BRILLIANCE: the call is written before the machine fails, and it goes to the technician already carrying the part.

An ATM technician with years in the field wanted the whole trade in one system: banks and fleet owners onboard their own machines, wear on every part raises the call before the fault, dispatch names the part and sends the call to the closest technician who has it in the van, the route lands the tight SLAs first, and the cash screen says which cassette runs dry on which day. Paid at 11:37am on a Saturday. Live at 2:29pm.

Paid 11:37am, live 2:29pm the same day Prediction without a sensor: wear off cycle counts Text dispatch, a rate card and a Crew tab added in one day
BRILLIANCE icon
BRILLIANCE
Anthony Davis · built by The App You Want
Fleet onboardingPredictive callsDispatch engineSLA routingCash forecastTechnician seat
Real screens
Dispatch board: 15 open calls, each with the machine, the branch, the technician it went to and the part it needs, tagged predicted, critical or past SLA
The boardEvery open call with the machine, the branch, who it went to and the part it needs. The fleet in these shots is a sample fleet built into the app.
A call: predicted by Brilliance, the machine and its four part health bars, the write up, bring CRDR-2210, sent to the closest technician with the part on board, proof photos
The callRaised by the app from wear on the machine's own cycle history. It names the part to bring and the technician it went to: closest, with that part on board.
Fleet: eleven machines with make, model, branch, status, the weakest part and its health, and days of cash left
The fleetEvery machine, its status, its weakest part and how many days of cash it has. Banks and fleet owners add their own.
Cash: every machine ranked by the day its cassettes empty, worked out from its own withdrawal history, with the vendor date
CashEvery machine ranked by the day its cassettes run dry, off its own withdrawal history. Nothing here touches a transaction.
A technician's calls: next up with one button, then everything assigned, each with the machine, the branch and the part it needs
The technician's callsWhat a tech sees: next up, one button, then everything assigned, with the part each one needs. Only their own calls.
Your route: five stops from the yard, miles, minutes out and time left on each SLA, ordered so the tight ones land first
The routeFive stops, 49.9 miles, about two hours of driving, ordered so the tight SLAs land first. Open the call or navigate from each stop.
See it live

BRILLIANCE is live

It installs from the browser on any phone, no app store. Seats are by invitation: a fleet administrator adds a person, and that person signs in with a code emailed to them. No passwords. One bank's terminals are never visible to another bank's technicians.

Open BRILLIANCE →
The challenge
"I want to take my knowledge along with futuristic technology and package it into a digital software solution."

The application came from a working ATM technician and ran to pages: predictive monitoring, an AI diagnostics layer, a dispatch engine, route optimization, voice close outs, a reporting dashboard, and a note that the whole thing should transfer to other industries later. The gameplan call was the following Tuesday. Then the offer opened with two straight answers. We are not building the transaction switch: the machines already run on a processor, so the app reads from it and never sits in the money path, and there is no PCI audit to pay for. And the internals on the big three manufacturers are locked, so live telemetry off the machine is not something anyone can promise. If a bank opens that data up, the system points at it and the monitoring switches on with nothing rebuilt.

Everything else went in the offer. The owner added AI driven and transferable, then a domain, a splash screen and a black and white theme, then predictive cash management. Paid at 11:37am on a Saturday. Live at 2:29pm.

What we built
An ATM managed services platform, from the bank's console to the technician's van.
▣

Fleets that onboard themselves

Banks and fleet owners add their own sites and machines and manage them on their own seat. Every tenant is walled off. Reports measured off the calls themselves: uptime, inside SLA, mean time to repair, caught before failing.

▲

The call before the fault

Every part on every machine carries a cycle count against its rated life. When one is close to done the app raises the call itself, writes it up, and names the part number to bring. No sensor, no manufacturer data.

→

Dispatch that knows the van

Each technician's van stock is in the app. A call goes to the closest technician who is carrying that part, by hand or by the engine overnight, and lands as a text with the reference, machine, location, part and SLA clock.

$

Cash, proof and the crew

Cash pickups scheduled off each machine's own withdrawal history. Photo proof on every call, from arrival to back in service. A Crew tab so a tech at a machine can reach whoever was on it last week. Voice close outs and an assistant on the owner's own AI account.

The signature feature
Dispatch will not send a technician to a job they cannot finish.

The write up names the part. The van screen says what each technician is carrying. So the engine does not pick the nearest person, it picks the nearest person with that part on board, and says so on the call: "closest with the part on board". A dispatch engine that always finds the part in every van is a dispatch engine nobody can see working, so the sample fleet's vans are stocked unevenly on purpose. The technician opens the app to one card, Next up, one button, and a route ordered so the tight SLAs land first, not the nearest stop.

Part named on the callVan stock per technicianClosest with the partTight SLAs first
How it moved
Two honest noes on Friday, a paid build Saturday morning, live by 2:29pm.
22 Aug
2:48pm, the application, pages of it, from a working ATM technician.
25 Aug
11:30am, the gameplan call.
28 Aug
1:07pm, the offer in writing, opening with the two things we will not build and why. 2:39pm: make it AI driven and transferable to other industries. Added.
29 Aug
8:05am: the domain, the splash screen wording, the black and white theme. 9:14am: predictive cash management added to the offer. 11:37am, paid. 2:29pm, live: fleet onboarding, predictive calls, dispatch, routing, the assistant, voice close outs, reports and cash.
1 Sep
The owner's questions answered in the build. 12:45pm: a dispatched call lands as a text, switched on per technician. 3:03pm: a rate card screen that prices a fleet you have not signed yet. 5:48pm: the Crew tab, technician to technician, inside the app so it costs nothing.
3 Sep
The first change list, sent the night before: a splash screen that holds seven seconds and dissolves into a home page in the owner's own words, with the feature list under the sign in. Built in the morning, photo and video proof on every call by the afternoon.
Want an app like this?
Tell us the app you want See it live