GuestGraph-powered AI Hospitality Command Center

Turn every table into an intelligent hospitality experience.

TableSage gives restaurants a multilingual AI concierge that recommends only from the approved menu, answers guest questions, routes staff requests, and turns dining-room intent into live operational insight.

Menu-grounded AI·Multilingual guest chat·Staff-routed requests·Allergen-aware guardrails·Real-time dashboard
Live Hospitality Pulse
Zafran · Saturday service
28
Active tables
14
Recos today
6
Open requests
3
Allergy reviews
ES
Table 12Menu-grounded
Vegetarian and a bit spicy
Matched 3 menu items
ES
RecommendationFeatured today
Cauliflower Shawarma · $17
Why: vegan, mild heat, chef-featured
EN
Table 7Allergy review
Nut allergy — is muhammara safe?
Allergen flag: walnuts
ES
Patio T10Resolved
Más agua por favor
Acknowledged · 2m 14s
Recommended action
Add allergen info to 7 menu items — top guest question tonight.
Menu-grounded
no hallucinated dishes
Allergen-aware
staff review surfaced
Multilingual
EN · ES · AR with RTL
Live-routed
Realtime to staff
The problem

QR menus solved printing. They did not solve hospitality.

Guests still have questions. Staff still get interrupted. Owners still cannot see what guests wanted but never ordered. A static menu does not guide, translate, explain, route, or learn.

Guests scroll instead of choosing

A long PDF menu is not hospitality. TableSage helps them ask naturally and get dish suggestions.

Dietary questions slow staff down

Vegan, halal, gluten-free, spice, allergen — answered from approved menu, routed to staff when it matters.

Menu data is incomplete

Allergen and dietary gaps surface in the owner dashboard before a guest finds them.

Staff requests get lost in waving

Every dish interest or call-for-help becomes a clean staff-visible request with table context.

Owners cannot see guest intent

Capture what guests asked, considered, and never ordered — not just what hit the POS.

GuestGraph

One live map of guests, menu, service, and intent.

GuestGraph connects menu data, guest questions, languages, recommendations, staff requests, and owner analytics. Not just answers — guest intent turned into service action and owner insight.

Node 01

Menu

Categories, dishes, prices, availability, tags, allergens, featured items.

Node 02

Guest intent

Language, table, dietary need, budget, spice, occasion, question type.

Node 03

AI Concierge

Grounded answer, recommendation cards, follow-up suggestions, safety warnings.

Node 04

Staff flow

Notify staff, allergy review, order interest, acknowledge, resolve, escalate.

Node 05

Resolution

Status, staff action, guest confirmation, table history.

Node 06

Analytics

Top dishes, top questions, dietary filters, menu gaps, conversion cues.

Menu → Guest Intent → AI Concierge → Recommendation → Staff Flow → Resolution → Analytics.

AI Concierge

A menu assistant that feels like hospitality, not helpdesk software.

Guests ask naturally, speak in their language, and receive grounded dish recommendations with clear prices, tags, and next actions.

  • Ask menu questions in natural language
  • Filter by dietary preference, spice level, budget
  • Ask about allergens — gets staff-review caution
  • Ask for popular or featured dishes
  • Ask in English, Spanish, or Arabic with RTL
  • Use text or voice (push-to-talk)
  • Send dish interest to staff in one tap
Z
Zafran · Table 7
Concierge · English
Menu-grounded
Something vegetarian and a bit spicy? My friend has a nut allergy.

Please confirm with staff before ordering — I can notify them now.

For you, the Cauliflower Shawarma is vegan, mildly spiced, and nut-free. Avoid the Muhammara — it contains walnuts.

Cauliflower Shawarma
Vegan · mild heat · nut-free
vegangluten-free
$17
Eggplant Fatteh
Vegetarian · gentle heat · contains sesame
vegetarianSesame
$18
More vegetarianSpicier mainsCall staff
Ask anything about the menu…
Menu-grounded AI

The AI only recommends what the restaurant actually serves.

No hallucinated dishes. No hidden allergen guesses. No unreviewed safety claims. If the menu does not say it, the AI does not pretend.

  1. Step 01
    Guest sends a message
  2. Step 02
    Server loads tenant-scoped menu + availability
  3. Step 03
    AI receives only approved menu items
  4. Step 04
    Structured output returns item IDs + reasons
  5. Step 05
    Server validates IDs — invalid ones discarded
  6. Step 06
    Allergy questions force staff review
  7. Step 07
    Staff request created on guest tap
  8. Step 08
    Analytics event stored for the owner view
Voice & multilingual

Let guests speak naturally — in their language.

Voice-first flows for guests who prefer speaking, tourists who need translation, and accessibility scenarios where typing is awkward.

Push-to-talk

Visible listening state, editable transcript before send. No always-listening.

EN · ES · AR (RTL)

Right-to-left support for Arabic, with proper script and direction.

Voice fallback

Voice always has a text alternative. Reduced motion respected.

Staff Service Flow

Guest intent should reach staff before frustration does.

Every AI conversation becomes a clean, staff-facing request with table, language, dish, urgency, and context.

Open requests
2 priority
T7
Allergy questionENAllergy review
Is the muhammara safe for tree nut allergy?
T7
Ready to orderEN
We are ready
T10
Call serverESResolved
Más agua por favor
  • Realtime queue subscribed to every new request
  • Allergy-sensitive items pinned with crimson + icon + text (never color-only)
  • Acknowledge · Resolve · Escalate · Needs manager
  • New-request chime and browser notification badge
  • Multilingual context shown alongside guest's actual words
Owner Command Center

See what the dining room is asking for.

Live sessions, request queue, top recommended dishes, menu completeness, and language mix — in one calm operational view.

28
Active tables
6
Open requests
3
Allergy reviews
Lamb Kofta
Top dish today
Vegetarian
Top intent
82%
Menu completeness
2m 14s
Avg response
EN 62% · ES 28% · AR 10%
Language mix
Menu Intelligence

Your menu should get smarter every service.

Missing tags, unclear descriptions, high-question items, and dietary-data gaps surfaced so owners can fix them between shifts.

Allergen completeness
82%
7 items missing data
Featured engagement
+34%
Lamb Kofta vs. last week
Lost-intent demand
11
Asked for but not orderable
Misunderstood dishes
3
Add prep notes
High-question dishes
5
Consider description rewrite
Upsell pairing cues
8
Mint Tea ↔ Baklava
Dining Room Signals

Capture the demand that never reaches the POS.

What guests ask, hesitate over, filter out, and request from staff — turned into operational intelligence.

Top guest intents · last 7 days
Vegetarian
78%
Spicy
64%
Gluten-free
41%
Featured
38%
Allergen
19%
Pairing
14%
Weekly owner brief
  • Top dish: Saffron Lamb Kofta (recommended 38× · ordered 31×)
  • Top question: vegetarian + spicy options
  • Allergy reviews: 7 (all confirmed by staff)
  • Menu gap: 7 items missing allergen data
  • Recommended feature: Halloumi & Date appetizer
  • Language trend: Spanish +18% week-over-week
Table Access Studio

Every table gets its own intelligent entry point.

Generate QR codes for each table, counter, room, or zone. Track guest sessions by location.

Table 7
Dining · active
Table 12
Dining · active
Patio 10
Patio · active
Bar B1
Bar · active
Human controls

AI can guide the guest. Staff still own the service.

TableSage keeps operational control with the restaurant and avoids overconfident answers where safety, availability, or staff judgment matters.

Menu-grounded only

AI never recommends what isn't on the live menu.

Server-side ID validation

Invalid recommendations are discarded before render.

Allergy review enforced

Allergy questions always route to staff.

Availability respected

Unavailable items hidden from recommendations.

Owner-visible AI activity

Discarded items and high-question dishes surfaced.

Demo isolated from prod

Demo seed data lives in its own restaurant tenant.

Your QR menu can finally answer back.

Launch the Zafran demo and see how guests ask questions, get menu-grounded recommendations, request staff, and give owners live insight into the dining room.