Service at the pass, a cellar that depletes as it pours, and menu engineering that protects the margin on every cover — 8 fine-dining solutions, narrated by Iris.
Covers paced to the kitchen's real capacity, allergens flagged before the plate leaves the pass, and each seating costed while it is still seated.
One mistimed course or one forgotten preference undoes months of reputation. Excellence at this level is memory plus timing — at a scale no brigade can hold in its head.
Guests known by name and preference — every visit, every allergy, every occasion, surfaced before they sit down.
Tasting menus paced to the minute — each course fired only when the previous one has been cleared.
Wine paired by margin as well as palate, every pour depleting the bin, every cover measured while it's still seated.
Covers, guest recognition, wine margin and pass timing — live, with Iris naming the detail that separated a good night from an exceptional one.
A POS built for how restaurants actually run — orders, split bills, any tender, receipt in seconds, while Iris suggests the right add-on and flags margin drift.
Course firing lives on the till. Staff fire each course from the POS — the kitchen never receives the next one until the table is ready.
It keeps selling when the line drops. Service continues through an internet outage and reconciles when the connection returns.
Iris prompts the server. The pairing, the supplement, the occasion upsell — surfaced on the item card, not left to memory.
Financial, customer, operational and AI-learning KPIs on one live board — with Iris narrating the one move that matters and the beat that recovers it.
Every module feeds one board. Pass timing, cellar margin, guest recognition and labour all post into the same view.
Forecasting, not just reporting. L2 models project where the service is heading while there's still time to change it.
Iris names the cause. Not the number that moved — the course, the section and the seating behind it.
The finance cockpit — trial balance, P&L, balance sheet, cash flow and sub-ledgers that tie out to the penny. Every sale from the till posts here in real time.
Sub-ledgers that tie to the penny. AR, AP and inventory reconcile themselves rather than being chased at close.
Deposits carried properly. Reservation deposits sit in escrow and net against the final bill without a manual journal.
Ask Iris for any number. Any outlet, any period — answered in the conversation, not a week after month-end.
Demand-planned replenishment across every outlet and distribution centre. Iris forecasts by daypart and weather so the right stock lands before you need it — not after.
The cover forecast drives the order. Expected covers by seating become prep quantities and purchase orders, not a chef's estimate.
The cellar is part of the chain. Bin levels and vintage availability feed replenishment the same way produce does.
Daypart and weather in the forecast. A rooftop service and a Tuesday tasting menu are not planned from the same curve.
Digital checklists, shift routines and standard operating procedures across every outlet — completion tracked live, exceptions escalated. Frontline execution you can actually see.
Mise en place, signed off. Opening and pass routines are completed in the system, not assumed to have happened.
Course timing rolls up. Per-course deviation aggregates by service, by section, by outlet.
Exceptions escalate themselves. The routine that didn't complete surfaces to a human; the rest stays quiet.
Rostering, attendance and labour percentage measured against forecast demand. Iris flags the overstaffed lulls and the understaffed peaks before the shift, not after.
Brigade built from the forecast. Section and pass coverage come from expected covers, not from last week's rota.
Labour measured against demand. Labour percentage is read against forecast covers, so a quiet Tuesday isn't judged like a full Saturday.
Flagged before the shift. An overstaffed lull or an understaffed peak is caught while the roster can still change.
Deterministic logic under the hood (the “Zen Rules” layer) — no model, no probability. Pre-orders, allergens and briefs are captured, checked and surfaced on schedule, every service.
Course 4 contains an ingredient flagged on a pre-order for this seating. Raised 24 hours ahead of service — while a substitution can still be planned.
Chef's table and tasting menu experiences require advance pre-orders for dietary restrictions, allergies and preferences. This module captures, organises and surfaces all pre-orders before service.
Guests submit dietary requirements and preferences at reservation time — before they arrive.
The chef receives a single pre-order sheet per service — all allergies and special requests in one view.
Pre-order allergens are checked against the planned menu before service — conflicts raised while there's still time to substitute.
Pre-order details link to specific courses, so substitutions are planned in advance rather than improvised.
Host and server receive a one-page guest brief before arrival — name, occasion, preferences, notes.
Regression, forecasting and attribution under the hood (the “Zen Models” layer) — course pacing, table availability, no-show risk and revenue per cover, predicted before they land. Four solutions run on this layer.
Multi-course tasting menus require precise pacing — each course fired to the kitchen only when the previous one has been cleared, with timing logged and enforced by the POS.
Staff fire each course from the POS — the kitchen never receives the next course until the table is ready.
A course-timing regression model benchmarks each course against historical averages and flags deviations in real time.
Time between each course is logged per table — showing exactly where the sequence ran long.
Wine pairings are linked to each course — the sommelier is prompted to serve before the food arrives.
When a course is ready for all guests, every station is notified to plate at once.
Iris predicts when each occupied table will become available — so the host can manage the waitlist and seat walk-in guests with precision, not guesswork.
Dwell time per table is shown live on the floor plan — how long each party has been seated.
A regression model trained on party size, day, service and course progress predicts availability to within ±8 minutes.
The predicted free time becomes a quoted wait for the waitlist — the host gives a number rather than an apology.
Tables significantly beyond predicted dwell are flagged, and a gentle intervention is prompted if needed.
A turnover efficiency score tracks the gap between one party leaving and the next being seated.
Fine dining reservations carry deposit requirements, special requests and occasion notes. The system manages the full lifecycle — from booking, to arrival, to post-dining follow-up.
Website and phone bookings land in the same system with full guest details.
A no-show predictor scores each booking 48 hours ahead, enabling a proactive confirmation call on high-risk reservations.
The cancellation policy window is applied automatically — deposit released or retained per your terms, with no negotiation at the desk.
Deposits are charged at booking, held until the reservation is honoured, then deducted from the final bill.
Birthday, anniversary, dietary requirement and preferred table are captured and surfaced at arrival.
Revenue per cover is the key metric in fine dining. This module tracks it live — by service, by day, by server — and benchmarks it against historical averages and targets.
RevPAC updates with every order placed — the current service is visible on the manager dashboard.
A driver-attribution model decomposes the variance — how much came from price, mix, volume or supplement attach.
RevPAC is compared across lunch, early and late seatings — so a soft second seating is visible against a strong first.
RevPAC by server identifies upselling stars and underperformers without subjectivity.
Target versus actual shows the live gap with time remaining in service, while it can still be closed.
Not a screenshot — the actual agent. Natural-language answers over the live service. Three solutions run on this layer.
A curated pairing guide built into the POS — sommelier or server can suggest the right wine for every dish, log the recommendation, and track what was accepted.
Every menu item carries pre-loaded pairings, visible to staff on the POS item card.
A recommendation model scores acceptance by guest profile and dish, surfacing the suggestion most likely to land.
Every suggestion is logged as accepted or declined — acceptance rate tracked per dish and per sommelier over time.
Wine is tracked by vintage, bin and bottles remaining, so staff know availability before recommending.
Open bottles deplete by the glass — the system knows when a bottle runs out without a physical count.
A recommendation engine calibrated for high-average-order-value fine dining — suggesting premium dishes, supplements and experiences at the optimal moment.
When a main is ordered, Iris suggests the truffle supplement, the wagyu upgrade or the premium preparation.
A collaborative-filtering model trains on acceptance by dish, guest profile and table spend trajectory.
When glass orders reach bottle economics, the server is prompted to offer the bottle — better value for the guest, better margin for the room.
Occasion-aware: a birthday table prompts a dessert cake or a champagne toast.
Every suggestion is logged accepted or declined — the model retrains on actual conversion, not intent.
Every visit, preference and note recorded against a guest profile. Return guests are recognised, preferences surfaced to the host, and their experience personalised before they sit down.
Dates, dishes, spend and server — a full history per guest, searchable by name or phone.
A churn classifier identifies guests who haven't returned within their normal frequency and triggers a win-back.
Birthdays, anniversaries and milestones are held against the profile and resurfaced on the next matching date.
Dietary restrictions, seating preferences and disliked ingredients are recorded once and available always.
At check-in the profile is surfaced — the host greets by name, with context. Post-visit ratings flag anything that needs follow-up.
No call to book, no deck to sit through — the till, the cellar, the pass and the guest's profile, running on live data and narrated by Iris.
Each solution is tuned to the venue types it actually fits — a chef's table and a rooftop are not paced from the same model.
You configure a sector playbook, not a custom project. Iris applies it herself — agentically, from day one.
Onboarding is agentic. Iris connects the POS, the reservation book and the cellar list directly — no manual data mapping.
The playbook applies itself. Sector thresholds go live immediately, then tune from real service data, room by room.
Live in weeks, learning from day one. 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.
“At the premium end, one mistimed course or one forgotten preference undoes months of reputation. Excellence at this level is memory plus timing — at a scale no brigade can hold in its head.”
Zentallio remembers for you: guests by name and preference, tasting menus paced to the minute, wine paired by margin as well as palate, every cover measured while it is still seated. Underneath sits the same platform every sector runs — one data spine from the till to the ledger, three layers of intelligence, and an agent that narrates every screen, flags what needs a human, and acts on the rest.
One mistimed course — the sequence runs long and the second seating never lands.
A preference or allergy forgotten between visits, undoing months of reputation.
Wine poured without the bin depleting — the cellar margin drifts and nobody sees it until stocktake.
Pre-orders, allergen conflicts and the guest brief are surfaced deterministically, 24 hours before service.
Course pacing, table availability and no-show risk are predicted before they land.
“What drove the change in revenue per cover?” — decomposed into price, mix, volume and attach, in seconds.
Every Zentallio sector runs the same three-layer intelligence, tuned to how that format actually loses margin.