Cordiant for Restaurants

The hyper-local AI website every Simphony restaurant should have.

Every guest request answered, fulfilled, and transacted — without a human in the middle.

Sara — the conversational character — runs reservations, ordering, pairing recommendations, gift cards, and private dining inline against Oracle Simphony. One branded URL per restaurant. Group license, deployed inside the brand's own Oracle Cloud Infrastructure tenant.

The Cordiant for Restaurants surface for The Artisan — burgundy header with the restaurant's name and tagline, a hero photograph of a plated salmon with quinoa, asparagus, and edible flowers, and Sara greeting a returning guest by name with anniversary recognition: Good evening, Alexandra. Welcome back to The Artisan — and happy anniversary. Below: three quick options — Guide Me Through the Evening, Start with Wine, I Know What I Want.

From the moment a guest discovers your restaurant to the moment the check is paid. One URL. One conversation.

Your restaurant has never had a digital identity. Until now.

Every restaurant has a kitchen, a dining room, a sommelier, a signature dish. Ask yourself: does your restaurant have a digital presence that's truly its own? Not a menu PDF buried inside the group's central site. Not an OpenTable listing where you're sorted by rating between fifty competitors. Not a DoorDash storefront where the customer is owned by the marketplace. Not a brochure with a phone number.

Restaurant groups have built reservation programs, loyalty programs, ordering programs — each running on a separate vendor, each charging a take-rate, each holding the customer relationship hostage. The restaurant itself — the place where the brand promise is actually delivered — has never had its own digital presence inside that ecosystem.

Beneath that surface, the lifecycle is fragmented across OpenTable, Tock, Resy, DoorDash, Uber Eats, Grubhub, Yelp, Tripadvisor, Toast online ordering, ChowNow, gift card processors, email programs, and SMS marketing tools. Each a different vendor. Each with its own chrome. Each with its own take-rate. Each with limits on what it can see about the guest. Every commerce moment leaks to a third party. The reservation, the order, the gift card, the loyalty — all transacted on someone else's surface, under someone else's brand, against someone else's economics.

Cordiant for Restaurants changes that. A single URL — yourrestaurant.com — where your guests can do everything: explore, reserve, order, pair, gift, enquire about private dining. Not some things. Everything.

The reframe

For the first time, Oracle Simphony isn't just your back-office POS. It's your guest-facing platform.

The shift

One AI website replaces the patchwork.

Cordiant for Restaurants replaces the brochure-style restaurant website plus the OpenTable/Tock/Resy/DoorDash/Yelp/Tripadvisor patchwork with a single conversational commerce surface that transacts directly against Oracle Simphony. Menus, reservations, ordering, private dining, gift cards, and follow-up — one dialog, one brand, one commerce engine.

Discovery and ordering as one conversation.

Guests don't browse a menu PDF, and then jump to a third-party ordering app. They ask for what they want — by dish, by dietary restriction, by mood, by occasion — and the right items render in the same surface, ready to add to a check that lives in Simphony. Discovery and commerce in one continuous scroll, not two screens stitched together.

Direct revenue, not take-rate revenue.

Reservations, pre-orders, gift cards, and takeout all transact through your commerce engine — not through someone else's marketplace. The guest is yours. The check is yours. The cover charge, the loyalty, the repeat visit — all yours.

Simphony is canonical.

Every menu read, every item added, every reservation booked, every gift card issued is a live call against Simphony. No cached menu that drifts from the POS. No reservation that lives outside the kitchen's view. Your system of record never disagrees with your guest's surface.

Private dining and catering as first-class flows.

A guest enquires about a wedding rehearsal, a corporate off-site, a Sunday anniversary. The conversation captures party size, date flexibility, dietary needs, budget, and occasion — and hands the structured enquiry to your events lead with everything already gathered. No forms, no back-and-forth.

Brand by architecture.

Every response — including every AI-generated recommendation from your menu and wine list — renders in your restaurant's brand. Tone, typography, imagery, layout. Inherited from a theme defined once and applied across every surface the guest touches.

In tenant. In your cloud.

Deploys inside your brand's OCI tenant. No cross-cloud hop between the commerce surface and the POS. Under your existing security posture. No new SaaS vendor in the data path. No guest data leaves your tenant. No marketplace intermediary reading your order flow.

Group-wide, property-consistent.

A restaurant group gets one platform across every concept — the steakhouse, the seafood house, the neighborhood bistro. Each renders in its own brand, shares the commerce engine and the architecture, and consolidates data at the group level where the CFO actually wants it.

Origin

Solved in the hardest environment first. Packaged for restaurants.

We built guest commerce for luxury hotels on Opera Cloud — a live reference deployment on the Oracle Hospitality Integration Platform sandbox. Three years of refinement on the most demanding commerce surface in hospitality.

The restaurant layer was always part of that stack. Guests at a hotel don't think in terms of "room system" and "dining system" — they ask about a table at Nobu in the same conversation they extend their stay, and the platform executes both. That restaurant layer has now been packaged into a dedicated platform for Simphony restaurants — available to restaurant groups and standalone restaurants who don't need the rooms side of the stack.

You get the same three-layer architecture, the same conversational-plus-deterministic execution model, the same brand-by-architecture rendering. Sized to a restaurant group's scale, priced for a restaurant group's margin, deployed against Simphony as the system of record.

Why it fits

A restaurant is the cleanest commerce surface there is.

Most categories don't fit conversational commerce well — too many SKUs, too few photos, too much retrieval, too little structured truth. Restaurants are different. Four properties of how a restaurant actually operates make conversational commerce a near-perfect fit:

Catalog in the prompt.

A single restaurant has fifty to three hundred items. A restaurant group is typically under two thousand five hundred across venues. The full catalog — with live inventory, live pricing, live modifiers — fits into every conversation. No retrieval layer, no vector store, no embedding drift. The model reasons against the whole menu.

One continuous scroll.

Discovery, choice, modification, reservation, check, and confirmation compose into a single vertical stream. No search page. No results page. No separate cart page. No separate confirmation page. Commerce happens in place. The guest never leaves the conversation.

Mobile as the native form factor.

Sixty to seventy percent of restaurant traffic is mobile. Typing searches on a phone is where the existing restaurant web loses its guests. A stream that responds to plain language — something light, dining in thirty minutes, for two — removes the search box entirely and replaces it with the conversation the guest would have with a maître d'.

Menu as canonical artifact.

A restaurant already maintains its menu as a live, structured artifact inside Simphony — the catalog is not an abstraction produced for the commerce surface, it is the kitchen's source of truth. The conversation reads the same menu the line cook is plating from. No second catalog. No data-modeling exercise to turn inventory into a shoppable taxonomy. The restaurant is the rare commerce category where the product database was already built, by operators, for themselves.

In practice

Four moments, one commerce engine.

Every interaction runs the same split — an intelligent conversational surface on top, deterministic application code underneath, executing against Simphony.

A reservation

"Table for four Saturday at 7:30 — my partner is gluten-free."

The platform checks Simphony availability, offers the closest times, confirms the booking, and attaches the dietary note to the cover record. The kitchen sees the gluten-free flag when the guest walks in. No separate reservation app. No data that has to be reconciled back to the POS the next morning.

An order

"Something lighter than the tasting — maybe the sea bass, and a glass of whatever the sommelier would pick."

The platform reads the live menu from Simphony, synthesizes a pairing from the wine list and the chef's notes, renders the dish and the glass in a branded card with the prices, and on confirm opens a tab in Simphony with the items already attached. The check lives in your POS from the moment the guest says yes.

A private dining enquiry

"We're booking our daughter's engagement dinner — about 18 people, a Saturday in October, she's allergic to shellfish."

The platform captures the party size, the date flexibility, the dietary restrictions, the occasion, and the guest's contact details — and hands the events lead a pre-structured enquiry with everything already gathered. The next conversation is a real proposal, not a form-filling exchange.

A gift card for a regular

"I want to send $500 to my parents for their anniversary — they come every year around this time."

The platform identifies the guest from their history, issues the gift card through your commerce engine, attaches a note, and notifies the property so the team can acknowledge the occasion when the parents arrive. Direct, personal, and reconciled into the same system that runs the rest of the night.

Sara, in motion

The full guest arc, in one continuous scroll.

A working build of Cordiant for Restaurants on a Simphony sandbox — from menu browse to sommelier-grade pairings, every moment renders inline against the restaurant's brand. No second app. No marketplace intermediary. Discovery, ordering, and recommendation in the same conversation.

The Cordiant for Restaurants menu drawer for The Artisan — Starters, Mains, Cocktails, and Desserts tabs across the top, with the Lobster Bisque card showing a hero photograph of the dish, the description Maine lobster, cognac cream, chervil, allergen tags for SHELLFISH and DAIRY, a price of twenty-two dollars, and a one-tap Add button.
Menus render with prices, allergens, photography, and one-tap Add. The catalog reads live from Simphony.
Sara recommending wine pairings inline at The Artisan — a glass of Sancerre Henri Bourgeois 2022 from the Loire Valley with crisp minerality lifting a Lobster Bisque, and a Caymus Vineyards Cabernet Sauvignon 2021 from Napa Valley as a bold, structured pairing for Wagyu Beef Tenderloin. Each wine renders as a card with vintage, region, tasting note, and bottle and glass pricing.
Pairings render inline. Sara reasons against the wine list and chef's notes — discovery and beverage commerce in the same continuous scroll.
Two golden dice resting on a polished marble counter, with a point-of-sale screen blurred in the background — a visual metaphor for the unacceptable risk of AI that guesses at menu, availability, and pricing.
The reliability test

AI that never touches Simphony.

Most restaurant AI guesses. It guesses the lobster is on tonight. It guesses the pinot is $96. It guesses the table for four at 7:30 is still open. It pulls from cached menus, stale snapshots, training sets — and delivers the answer with absolute confidence.

Until the guest arrives and the lobster is 86'd. The wine list updated yesterday. The 7:30 table was given away an hour ago. In any other industry, that's a bug. In hospitality, that's a 1-star review that mentions your restaurant by name.

Conversation and transaction are two different problems. Asking "what's good tonight?" is a conversation. Saying "I'll have the sea bass" is a transaction. One should feel personal. The other must be deterministic. Most restaurant AI treats them as the same problem. They're not.

Cordiant splits them in two. A cognitive layer that knows your guests — remembers preferences, understands occasion, makes every interaction feel personal. An execution layer that guarantees accuracy — the live menu from Simphony, the live availability, the deterministic check-write through Simphony's existing role-based access controls.

The AI cannot hallucinate an order. It cannot add a menu item that isn't on the menu. It cannot apply pricing the kitchen hasn't approved. It cannot commit a reservation outside service policy. Not because we trust it not to — because, structurally, it is not the thing executing the action.

Simphony stays canonical. Every check is direct. Every reservation is logged. Every modifier is guaranteed accurate. One source of truth.

The reliability test

Two questions for any AI vendor:

  1. When a guest says "I'll have the sea bass," is that order deterministic or probabilistic?
  2. If Simphony goes down for thirty seconds, what does your AI tell the guest?

If the order is probabilistic — your POS is at the mercy of a language model. If the AI keeps going when Simphony is down — it's making things up, with no source of truth.

100% reliable doesn't mean 100% available.
It means 100% honest.

Every answer sourced. Every transaction deterministic. Every check auditable. Every time.

Restaurant software should not act like the roll of a dice.

Deployment

Group first. Restaurants next. Configuration, not bespoke engineering.

Adoption is a configuration exercise against a platform that already works. The first restaurant in a group reaches production in six to eight weeks — brand theme, Simphony integration, identity federation, commerce engine, menu and knowledge ingestion, and payment rails, all configured against a platform that already works. Subsequent restaurants compress to two to four weeks because the group-level foundation is already live.

Cordiant for Restaurants deploys inside your brand's OCI tenant, under your existing security and compliance posture. No cross-cloud hop between the commerce surface and the POS. No new SaaS vendor in the data path. No guest data leaves your tenant. Identity federates through OCI IAM. LLM inference runs against the model provider you govern, under your data-residency and model-governance rules.

Two operating modes. One architecture. Self-operated: your technology team runs the platform, with Cordiant Group Support covering architecture, upgrades, and issue triage. Cordiant-operated: Cordiant runs the platform inside your tenant under an ops contract, with audited access and no data leaving your cloud. Tenant ownership, identity federation, and the trust posture are identical in both modes. Only the operator changes.

6–8 weeks
First restaurant from kickoff to production
2–4 weeks
Each subsequent restaurant in the group
In tenant
Your cloud. Your audit trail. Your control.
Commercial model

Perpetual group license. Per-restaurant implementation. No take-rate.

The same predictable structure as Cordiant for Hotels, sized to restaurant-group scale and margin. Predictable line items. No revenue share. No per-transaction metering. No marketplace take-rate. Commercial patterns your CFO already understands from POS, reservation, and loyalty contracts.

Group license
Group-sized
Per-restaurant implementation
Flat per unit
Per-restaurant annual support
Flat per unit / year
Specific terms are sized to group scale and shared under NDA during a briefing. We price to align with restaurant-group economics — not to replicate a hotel-brand fee. The structure above is the shape; the numbers are the conversation.
Next step

From conversation to first restaurant in production.

Thirty to forty-five minutes with Cordiant leadership. We walk the architecture, show a working build of Sara on the Simphony sandbox, and propose a first-restaurant scope. Every guest request answered, fulfilled, and transacted — without a human in the middle. Every transaction sourced. Every check auditable. Every time.

Stop describing the restaurant. Start running the commerce.

Schedule a Briefing