An executive briefing on how Iris — Zentallio's AI decision agent — runs the counter, the kiosk, the drive-through and every aggregator, minute by minute.
Lane depth, order-to-window time, and the second window — measured to the second, because that's where the rush is won or lost.
One order book, sequenced by promise time — not by whoever shouted loudest.
Kiosk, drive-through, counter and every aggregator on one order book — sequenced by promise time.
Lane depth, order-to-window time and the second window measured to the second — the queue is the P&L.
One menu across kiosk, app, web and five aggregators. An item goes 86 at the counter and disappears everywhere in the same second.
Everything below reads from — and writes back to — the same order book, in real time.
The terminal a cashier runs all day. Order entry, payment and Iris's own upsell prompt, all on the same screen the till already has.
One screen, order to payment. Order entry, ticket and card payment happen without switching apps or terminals.
Same order book as everywhere else. Whatever's rung up here reconciles automatically with kiosk, app and aggregator sales.
Iris prompts the cashier, too. A combo upgrade or add-on surfaces right on the till, timed to what's already in the order.
The live financial view every other product feeds into — net margin, food cost and forecast accuracy, reconciled to one number.
Every module feeds one ledger. Till, kitchen, workforce and supply chain all post to the same financial view.
Forecasting, not just reporting. L2 models project where the quarter is heading, not only where it's been.
One number, not four. "Food cost" means the same thing whether Iris, a GM or the CFO is asking.
The full ledger — P&L, balance sheet, cash — reconciled down to the store, not just the region.
Every sale posts directly. Sales, refunds and voids land in the ledger without a manual close.
Sub-ledgers reconcile themselves. AR, AP and inventory tie out automatically, store by store.
Ask Iris for any number. Any store, any period, instantly — not a week after month-end.
Connects every vendor, delivery and ingredient movement into one supply-chain view — from the receiving dock to the plate.
Every delivery logs itself. Receipts and transfers are captured automatically, not written on a clipboard.
Vendor pricing, centralized. Cost and lead times are tracked across the network, not negotiated per store.
Feeds Iris directly. Recipe-Based Inventory Depletion runs on Nexus data, so drift is caught same-day.
The operations layer — service times, staffing adherence, compliance — watching every store in real time.
Service times, tracked store by store. Order-to-window and counter time roll up across the whole network.
Surfaces what needs attention. Not a report on all 142 stores — the two or three that actually need a look today.
Feeds Zen Rules and Zen Models. The operational data Iris's other two layers run on comes from here.
Workforce and scheduling — shift plans built against the demand curve, with labor compliance guardrails built in.
Staffing follows the forecast. Shift plans are generated straight from Zen Models' hourly demand curve.
Compliance is built in. Break rules and maximum hours are guardrails, not a separate checklist.
Flags bad schedules early. An over- or under-staffed shift is caught before it's published, not after the rush.
Deterministic thresholds under the hood (the “Zen Rules” layer) — catch a known problem the instant it crosses a line. Six solutions run on this layer.
Store #12 — Zinger Burger sold out at the counter. Pulled from kiosk, app, web and all 5 aggregators automatically.
Every ticket fires the instant an order is placed and routes to fry, grill or assembly by build time — not by whoever's screen is free.
Ticket fires the instant the order is confirmed, on any channel.
Routed to the station with the shortest build time, not the nearest screen.
Re-sequences automatically the moment a station backs up.
Breakfast closes and lunch opens on schedule — across kiosk, app, web and every aggregator, in the same second.
Daypart windows configured once, per store or per group.
Kiosk, app, web and all aggregators switch simultaneously.
No manager has to remember to flip the menu by hand.
The moment an item sells out at the counter, it disappears from every channel — before a guest can order what you don't have.
Counter marks an item sold out.
Rule fires across kiosk, app, web and every aggregator.
Item reappears automatically the moment it's restocked.
Menu, price and availability held as one source of truth — every aggregator reads the same feed, so nothing drifts out of sync.
One menu feed powers kiosk, app and every aggregator.
A price or availability change lands everywhere at once.
No re-uploads, no per-platform spreadsheet.
One rule set, defined once at HQ — a new store inherits the standard on day one, not after a training cycle.
Rules defined centrally, versioned like code.
New stores inherit the current standard automatically.
Local exceptions require sign-off, not silent drift.
Fridge, freezer and holding temperatures monitored continuously — a breach alerts before spec is lost, not after a health inspection.
IoT sensors log every fridge, freezer and holding unit.
A threshold breach triggers an alert immediately.
A compliant audit log is generated automatically.
LightGBM / GBT forecasting under the hood (the “Zen Models” layer) — predicts demand, prep and staffing before the rush starts. Four solutions run on this layer.
Schedules people against the hourly demand curve — not against habit, not against last week's rota.
Forecasts order volume by hour, by store.
Suggests a staffing plan matched to the curve.
Flags shifts that are over- or under-scheduled before they happen.
Forecasts where the next lane or counter bottleneck forms — before it forms, not in next week's report.
Tracks order-to-window and counter time, live.
Learns the pattern behind recurring bottlenecks.
Flags a forming delay while there's still time to act.
Watches every car in the lane, predicting order-to-window time before it slips past promise.
Tracks lane depth and per-car dwell time.
Predicts when the second window will miss promise time.
Suggests a fix — a second operator, a paused promo — before it does.
Flags when ingredient usage drifts from the recipe — shrink caught the day it happens, not at month-end.
Every sale depletes inventory by the exact recipe.
Actual usage is compared to expected, continuously.
A drift beyond threshold raises an anomaly, same day.
Not a screenshot — the actual agent. Natural-language answers over the live order book. Three solutions run on this layer.
Suggests the next item a guest is most likely to add, tuned to time of day and order history.
Reads the current order and the daypart.
Ranks the next-best add from what converts, not a fixed script.
Learns from what guests actually accept or skip.
Builds the combo and modifier prompt most likely to convert — tuned per channel, kiosk to counter to app.
Matches combo prompts to the item just ordered.
Tunes the offer differently for kiosk, counter and app.
Retires prompts that don't convert, automatically.
Sends the right offer to the right guest, timed to when they're most likely to come back through the door.
Segments guests by real order history, not broad demographics.
Times each offer to typical return windows.
Retires offers that guests stop responding to.
No professional-services project, no six-month build. Iris configures the sector playbook herself — agentically, from day one.
Onboarding is agentic. Iris connects the POS, kiosk and every aggregator feed directly — no manual data mapping.
The playbook applies itself. Sector thresholds go live immediately, then tune from real data, store by store.
Weeks, not quarters. There's no bespoke build to wait on — go-live is a configuration, not a project plan.
The first line of support is agentic — Iris resolves most of it herself. Our engineers pick up from there.
Layer 1 — Iris, 24/7. Configuration questions, anomalies and routine issues resolved directly, instantly.
Layer 2 — our engineers. Anything Iris can't close escalates automatically to a Zentallio engineer.
No blank tickets. Every escalation arrives with Iris's own diagnosis — engineers start from an answer.
"The margin lives in the rush: a lane that stalls, a fryer that runs ahead of demand, a promo that leaks — each costs more than a bad month of rent. Most operators only find out after close."
Zentallio runs the counter minute by minute: Drive-Through Management watches every car, the Kitchen Display routes each item to the right station, recipe-level depletion catches shrink the moment it happens, and the Staffing AI schedules people against the demand curve — not against habit. Configured as a sector playbook, not a custom project: live in weeks, learning from day one.
Small portion and recipe drifts are invisible at 412 orders/hr — until they compound.
A stale menu across five aggregators sells items you no longer have.
Peak-hour staffing guessed, not planned against the demand curve.
Recipe-based depletion and out-of-stock rules fire the instant a threshold crosses.
Hourly demand forecasts drive prep and staffing before the rush hits.
"Why did order-to-window spike?" — answered in seconds, not a week.
Every Zentallio sector runs the same three-layer intelligence, tuned to how that format actually loses margin.