An executive briefing on how Iris — Zentallio's AI decision agent — runs the floor plan, the coursing, the servers and every split bill, table by table.
Table turn time is the number that decides a Saturday. Iris seats, fires and paces against it rather than hoping.
Casual dining wins on turns and dies on friction: slow coursing, split-bill chaos, comps nobody can explain.
A thousand signals a night — seating, firing, comps, voids — and most of them are lost by morning.
Starters, mains and desserts fired to arrive together for the table — not for the ticket.
Every comp and void is attributed, and Iris names which section is drifting before the guests feel it.
One data spine — from the till to the ledger. Everything below reads from, and writes back to, the same floor.
Orders, modifiers, stock depleting as you sell — and Iris upselling every ticket without adding a second to the queue.
Order, course and settle on one screen. Server takes the order, courses it to the kitchen and splits the bill without switching terminals.
Stock depletes as you sell. Every modifier rung up moves inventory in the same second — no end-of-night reconciliation.
Iris prompts the server, too. A course upgrade or add-on surfaces on the till, timed to what's already on the table.
Covers, turn time, comps and course accuracy — live, with Iris naming the section where the evening actually slipped.
Four perspectives, one board. Financial, customer, operational and learning — every number live, not compiled next week.
Every module feeds it. Floor plan, kitchen, workforce and supply chain all post into the same view.
Iris names the cause. Not just the number that moved — the section, the shift and the server behind it.
P&L, balance sheet, cash and close on a single balanced ledger — a month-end that is a review rather than a rebuild.
Sub-ledgers that reconcile. AR, AP and inventory tie out automatically — not chased down at close.
Comps and voids carry attribution. Every one posts with the server, section and reason attached to the entry.
Ask Iris for any number. Any site, any period — answered in the conversation, not a week after month-end.
Forecast to prep plan to purchase order to the cold-chain van — eighteen modules, one supply chain.
The cover forecast drives the order. Expected covers by hour become prep quantities and purchase orders, not a manager's guess.
Vendor pricing, centralized. Cost and lead times tracked across the group rather than negotiated site by site.
Cold chain included. The van is part of the plan — eighteen modules from forecast to delivery.
The operations layer — surfacing exceptions and incidents across the estate, and leaving the rest alone.
Course timing, tracked per table. Time between courses rolls up by section, shift and site.
Surfaces what needs a human. Not a report on every site — the sections that actually slipped tonight.
Feeds Zen Rules and Zen Models. The operational data Iris's other two layers run on comes from here.
Rostering built from the forecast — attendance, payroll, working-time compliance, and the retention risk hiding behind the labour ratio.
Rosters follow the cover forecast. Section coverage is built from expected covers by hour, not from last Saturday's rota.
Compliance is a guardrail. Working-time rules and breaks are enforced in the roster, not audited after payroll.
Retention risk, surfaced early. The pattern behind the labour ratio — before a good server hands in notice.
Deterministic thresholds under the hood (the “Zen Rules” layer) — catch a known problem the instant it crosses a line. Split billing and open tabs run on this layer.
Table 14 — running tab has crossed the configured threshold. Surfaced to the duty manager while the table is still seated, not at settlement.
A table's bill can be split any way the guests choose — equally per head, by individual items ordered, or a custom combination. No mental arithmetic for staff.
Equal split — total divided by number of guests, each paying their share by card or cash.
Item-by-item — each guest selects their own items and pays exactly what they ordered.
Multi-tender per split — Guest A pays cash, Guest B taps card, on the same bill.
Service charge distributed proportionally across splits — no manual calculation.
A tab opened when guests are seated stays open until they ask for the bill. Staff add items at any time without re-opening an order.
Tab is created automatically when a table is assigned — no separate step.
Additional orders join the running tab — no new order, no duplication.
A high-spend tab approaching a configurable threshold is flagged for manager awareness.
Tabs transfer between tables or to the bar, and accept partial payment mid-service.
Forecasting and regression under the hood (the “Zen Models” layer) — predicts table availability, covers and no-shows before the service starts. Four solutions run on this layer.
A live digital floor plan showing every table's status — occupied, free, reserved, or being cleared. The host always knows where to seat the next guest without checking physically.
Drag-and-drop floor map matches the actual layout, updating as orders are placed and bills paid.
The turn predictor estimates when each occupied table frees up — availability seen 15 minutes ahead.
Colour-coded at a glance: free, occupied, reserved, cleaning — on any device.
Indoor, terrace and private dining on one screen; tables merge for large parties and split for separate bills.
Tables assigned to specific servers. Orders coursed to the kitchen in sequence — starter, main, dessert — the kitchen receives each course only when the previous one is cleared.
Every table is assigned to a named server — sales, tips and performance tracked per server.
Starter fires at order time; the main fires when staff course it. The kitchen never gets ahead.
The KDS shows course number and sequence — no confusion between tables' starters and mains.
The scorer tracks AOV, upsell rate and tip % per server — coaching opportunities without bias.
Phone and walk-in reservations captured in one system — with deposit tracking, no-show flags, and a live waitlist for busy periods. The host is always in control.
Bookings taken by phone or online — date, time, covers, special requests and contact recorded.
The no-show predictor scores each reservation 0–100, allowing an intelligent overbooking buffer.
Deposits are linked to the booking for large parties and peak nights, then deducted at billing.
The cover forecast by hour lets the kitchen plan prep quantities in advance.
Tip prompts on the card terminal at checkout — with staff tip allocation, pooling options, and full reporting. Service charge applied per your policy and distributed correctly.
Configurable tip options appear on the terminal at checkout — 10%, 15%, 20% or custom.
Cash tips are logged against the server and included in their performance report.
Fixed-percentage service charge is added automatically, with exempt categories excluded.
Tips per server, per shift, per week — a full audit trail for payroll integration.
Not a screenshot — the actual agent. Natural-language answers over the live floor.
The guest, remembered. Dietary restrictions and preferences are held against the guest profile and surfaced to the host and server before they reach the table.
The profile is surfaced at check-in — before the host or server approaches the table.
Guests are clustered by preference pattern, not by broad demographics.
Recommendations respect the restriction on record rather than relying on the server remembering.
Every card opens a live guided demo. Seven are detailed in this briefing; the remainder run on the same three layers.
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 floor plan and the reservation book directly — no manual data mapping.
The playbook applies itself. Sector thresholds go live immediately, then tune from real service data, site by site.
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.
“Coursing runs itself. Managers finally watch guests, not tickets.” GENERAL MANAGER · CASUAL DINING GROUP
Zentallio keeps the floor moving: table-turn prediction paces seating, coursing runs itself, every comp and void is attributed, and Iris tells the manager which section is drifting before the guests feel it. 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.
Slow coursing — the kitchen runs ahead of the table and the turn stretches past plan.
Split-bill chaos at settlement, with staff doing arithmetic while the next party waits.
Comps and voids nobody can explain — and a no-show table nobody backfilled.
Split rules and tab thresholds fire deterministically — at bill request, at seating, every time.
Turn prediction paces seating and the no-show score sets the overbooking buffer.
“Which section is drifting tonight?” — answered in seconds, while service is still running.
Every Zentallio sector runs the same three-layer intelligence, tuned to how that format actually loses margin.