From the marketing brief to the kiosk screen.
Cornac is a fleet-driving platform. It writes the campaign brief, enters the configuration into the POS, reads it back, drives the kiosk the way a customer would, and compares every location against the others.
Eleven modules, one chain — and one configuration per fleet, per POS vendor, per channel.
Five moves, one source of truth.
Between the marketing decision and the screen the customer sees, there is a brief written by hand, a configuration typed by hand, and nobody to read it back. Cornac holds all five moves end to end.
The brief
ScribeThe campaign brief is generated from the existing working files, published early, and signed off online.
The configuration
FeederItems, prices, recipes and kits are written into the POS under two-factor lock, instead of being retyped location by location.
The read-back
AuditorThe POS is read back and compared against the brief. What differs surfaces at entry time, not the night before launch.
The screen
DriverAn automation drives the kiosk like a customer and checks what is actually displayed, with the capture to prove it.
The fleet
CoherenceEvery location is compared against the others, read-only, without taking a single device out of service.
The Ledger is what ties it together — the file of record produced by the Auditor. The Driver reads it to know what to check on screen; the delivery and loyalty tests read it too. One campaign, described once.
Eleven modules, one name per job.
The family follows a clear convention: an agent's name that says what it does. Feeder feeds, Auditor reads back, Driver drives, Studio houses, Scribe writes, the Ledger records. You take what you need, in the order you need it.
Cornac Studio
it housesIn serviceThe workbench. The web front end and server that carry Feeder, Auditor, Coherence, Compliance and the fleet directory.
- A simulated mode lets you train and prepare a campaign without touching the POS
- Installs on a dedicated workstation, with no internet access
Feeder
it entersIn serviceReads the campaign brief in spreadsheet form and writes the configuration into the POS, in place of manual entry.
- Items, prices, recipes, kits, effective dates
- Every real write requires a password and a one-time code
Auditor
it reads backIn serviceReads the POS back and compares it against the brief. The gap surfaces at entry time, not in testing the night before.
- Price gap, missing item, diverging label, missing tax
- Produces the Ledger, the source of truth for every downstream test
the Ledger
it recordsIn serviceA campaign's file of record: what was asked for, what was written, what was read back.
- Read by the Driver, by delivery tests and by loyalty tests
- One campaign described once, for every test
Fleet Coherence
it comparesIn serviceCompares every location against the others, read-only, in one minute.
- Four axes: menu, discounts, tender media, service charges
- Three severities, each resting on a fact rather than on a resemblance between labels
Cornac Driver
it drivesIn serviceThe automation that drives the kiosk at the display level and checks what the customer actually sees.
- Full journey through to payment, with on-screen proof
- Nothing is installed on the device in service
Compliance
it holds the ruleIn serviceCampaign rules, written in plain language, checked against the real configuration.
- A rule can be read and argued without opening a technical file
- The same rules serve both the Driver and the Studio
Xray Sync
it filesIn serviceEvery run leaves a trace in the company's test repository.
- Jira and Xray: test, execution, environment, evidence attached
- The report is not a file left on someone's desktop
Scheduler
it triggersIn developmentTests run overnight, and on every delivery of a new application version.
- Configurable schedule and test sets
- Tells FAILED — the application has a defect — apart from NOT RUN — the environment was not ready
Cornac Scribe
it writesIn developmentThe campaign brief is no longer written by hand: it is generated, published early and signed off online.
- Completeness checks at generation: item with no price, no artwork, out of scope
- Sign-offs collected item by item, with author and timestamp
Activity Recorder
it remembersIn developmentBuilds the hourly history that POS systems never archive — the raw material of a forecast.
- An hour with no reading returns “unknown”, never “zero”
- The hour of record is the POS's, not the reading's
The brief is no longer written. It is generated.
Not spending four hours typing 1,000 items at six in the morning for a campaign that starts at ten.Programme director, quick-service restaurant chain
Today the brief is written by hand from working files, then dropped somewhere with its artwork. The decisions, meanwhile, arrive as they come — usually after the brief has gone out. Scribe reverses the order: the brief lands early, so the calls are already made on the day.
Generate
The brief is produced from the working files you already have: items added and removed, location scope, prices, effective dates. Completeness checks run at generation time.
Publish early
The brief goes online the moment it is produced, with a history of what changed between versions, and when. Leadership decides sooner.
Sign off online
Sign-offs and comments are collected item by item, with author and timestamp. Progress reads at a glance, and outstanding items get chased.
One product, six axes of configuration.
Cornac is not a script written for one fleet. What changes from one customer to the next lives in connectors, rules and scenarios — the engine itself does not move. That is what makes a new fleet a matter of weeks rather than months.
What the first read-through found.
Measured across the full estate of a quick-service restaurant chain, read-only, without taking a single device out of service. These are findings, not sales estimates.
Four items with no price at all
Present on one location's menu, therefore unsellable — the day before their national launch. Five and six euros everywhere else.
Fifteen items correct à la carte
Only their meal-deal price diverged, across eight locations. À la carte everything was right — which is precisely why nobody had spotted it.
A kids' menu at two prices
€4.99 here, €3.99 there, across eleven locations. A deliberate pricing zone or an oversight — only the operator can tell, and that is exactly the point.
Three labels for one discount
Same amount, three different wordings. And that wording is what prints on the receipt the customer walks out with.
What holds the product up.
Two engines, one vocabulary
The Studio carries configuration reads and writes; the Driver carries screen automation. They share the Ledger and the campaign rules — never two versions of the same truth.
A simulated mode that stays
Train a team and prepare a campaign without touching production, or even holding access to it. That is a decision, not a leftover.
A write lock in the right place
The lock sits where the calls pass, not in the interface — you do not get around it by opening another page. Without explicit setup, no write is possible at all.
An install that asks for nothing
The package is dropped and run: no dependency install, no outbound access, no administrator account on the production machine.
A trace, not a file
Readable reports per scenario and per campaign, filed into the company's test repository with the evidence attached to the run.
Rules written in plain language
A campaign rule can be read, argued and corrected without opening a technical file. That is what keeps it reviewable by the people who set it.
It says what differs. It does not say what is wrong.
A local discount is entitled to exist; so is a regional price. Only the operator can tell intent from oversight — Cornac's job is to put the gap in front of them before opening, not to decide for them. The word “anomaly” appears nowhere in its reports.
It does not replace a tester's judgement. It reads back what is comparable — prices, presence, labels, consistency across locations — and gives teams back the time to chase what no script will ever find.
Finally, it only sees what the interfaces expose. Allergens, for instance, are not in the POS: across a fleet's 756 items, not one references them. That check happens on screen, and we say so rather than let anyone believe otherwise.
The first fleet read-through lands in two to three weeks.
Access
Your POS vendor's application interfaces, read-only. That is almost always the real lead time on the project, not the engineering.
The fleet read-through
Read-only, with nothing taken out of service. A first pass over a fleet that has never been compared always finds something.
The kiosk, then the Feeder
Screen automation comes next, and writing comes last — because you only write once trust is established.
What people ask first.
Does Cornac work with our POS?
Do we have to install anything on our kiosks?
Does Cornac write into our production?
From how many locations is it worth it?
Where is the product hosted?
Who owns Cornac?
The mahout does not replace the elephant. He drives it.
A fleet has an elephant's strength and memory: it does exactly what it was taught, everywhere, without ever needing to be told twice. That obedience is also its flaw — it applies a mistake as faithfully as it applies a correct instruction.
Hence the mahout's job: point it where to go, and check that it went. Neither a trainer nor a passenger.
Cornac · by TTCA demo belongs on a real fleet, not on screenshots.
Send us two past campaigns and read-only access. We run the first fleet read-through in front of you.