Skip to content
Parla con le vendite Contatto Accedi
RestioX AI Suite

The AI Suite That Reads Your Restaurant’s Own Numbers

Eighteen AI tools built into the RestioX admin. They answer from your order lines, your recipe costs, your stock and your floor state — not from a general idea of what restaurants are like. Each one proposes and explains; you decide.

RestioX Admin — Restaurant Brain
Friday 6:12 pm — Riverside branch
Covers vs last Friday−11%
Butter Chicken now sells below costLeak
Table 7 waiting 24 min on mainsWatch
Roster 2 extra hands, 7–9 pmPlan
14 items sold out since WednesdayHide?
Projected tonight $4,180
Dismiss
Approve
You ApproveNothing runs unasked
18 ToolsOne shared context

18

AI tools inside the admin — no second app, no extra login

30-day

Rolling window of your own paid orders behind every answer

6

Kinds of margin leak the radar checks, each costed per week

100%

Of automated actions are a draft or fully reversible

Grounded in your data

It answers from your books, not from the internet

A general chatbot knows what a restaurant is. It does not know that your butter chicken costs 4.10 to plate, that Thursdays have run quiet since the road closed, or that the fryer station is a person short after eight. Before a word is generated, RestioX assembles a context pack from your own tables and asks the model to reason over that — and only that.

  • Paid orders and order items from the last 30 days, scoped to the branch you are asking about
  • Live menu items with recipe cost, so margin is calculated rather than guessed
  • Stock levels, wastage and expiry dates straight from the inventory module
  • Table state, open service requests and reservation load for the shift in question
  • Customer personal data can be scrubbed from every prompt before it leaves your server
  • A per-restaurant monthly token budget, checked before every call, that fails closed

That pack is also what makes an answer checkable. When Menu Engineer calls a dish a plowhorse, it is your order items and your recipe cost saying so, and both sit on screen beside the recommendation.

Reads from
Context pack — assembled server-side
Read for this answer
Paid orders, last 30 days4,812 lines
Menu items with recipe cost148
Stock items + expiry dates96
Tables + open service requestsLive
Rota, Fri 18:00–23:0011 staff
Before it leaves the server
Guest names and phone numbersScrubbed
Monthly token budget62% used
Branch scopeRiverside only
Fails ClosedBudget gate, every call
Marquee capabilities

Six of them you will use on a normal Tuesday

  • Restaurant Brain

    Ask in plain language — which starters lost money last week, why March Fridays were down, what the average check is on delivery versus dine-in — and the answer is assembled from your own order lines, with the figures it used shown beside it. What it learns goes into Restaurant Memory, so the next conversation starts where the last one finished.

  • Auto-Pilot

    It watches for situations worth acting on and proposes one concrete step: hide the six items sold out for two days, draft a win-back campaign for lapsed regulars, flag a dish now selling below cost. Nothing runs until you approve it, every approved action is a draft or a reversible change, and a fired action stays quiet for fourteen days so you are never nagged twice.

  • Menu Engineer

    Plots every dish on the margin-versus-popularity matrix — stars, plowhorses, puzzles and dogs — and says what to do with each one. Repricing suggestions apply to the live menu inline.

  • Profit Leak

    Six detectors run over your own figures: over-discounting, dishes sold below cost, drawers that do not balance, stock about to expire, refund spikes, and popular items on a thin margin. Each leak is costed per week and ranked.

  • Tomorrow Plan

    A next-day brief from the same weekday’s history: expected covers and revenue, the hourly demand curve and its peak, how many people to roster, what to prep, and what is likely to go wrong.

  • Live Ops

    The service floor as it happens — which tables are waiting, where pressure is building, and what the kitchen is about to fall behind on.

The full suite

Eighteen tools, one shared context

Each is a real page in the admin, not a prompt box with a new label — and they all read the same context pack, so a finding in one lines up with the numbers in the next.

  • Restaurant BrainAsk questions about your own restaurant in plain language.
  • Auto-PilotProposes concrete next actions; you approve, dismiss or undo.
  • Menu EngineerSorts dishes into stars, plowhorses, puzzles and dogs.
  • Dish DoctorDiagnoses why one dish is underperforming and what to change.
  • Profit LeakFinds and costs the six ways margin quietly disappears.
  • Live OpsReal-time read on the service floor while service is running.
  • Kitchen CommanderThroughput and prep guidance for the pass during a rush.
  • Tomorrow PlanNext-day covers, staffing and prep brief with a demand curve.
  • Photo CoachCritiques a menu photo and says exactly how to reshoot it.
  • Taste ProfileBuilds a taste model per returning guest from order history.
  • Personalized MenuReorders the menu a guest sees around what they actually order.
  • Table MoodScores every seated table 0–100 on how the visit is going.
  • SimulatorModel a price or staffing change before you commit to it.
  • Competitor SpyBenchmarks your dishes against nearby competitors you add.
  • Channel PositioningHow to price and position per sales channel you sell on.
  • One-Click CampaignDrafts a whole campaign — audience, offer, subject line, copy.
  • Restaurant MemoryDurable notes the AI writes to and reads back weeks later.
  • Voice AssistantAsk and instruct by voice while your hands are busy.
And three things that are not pages

A morning brief waiting on the dashboard when you sign in. A Cmd+K palette that opens any tool from any screen. An AI Control Center where you set autonomy per action, review the audit trail and watch token spend against your plan.

See it read your first week of orders

Load a menu, run a few services, then ask the Brain what it noticed.

Start Free Trial
AI Menu Upload

Photograph the menu you already have

Getting an existing menu into a new system is the job that stops most restaurants switching. Point a phone at your printed menu, or drop in the PDF your designer sent, and the extractor reads it back as a structured preview: categories, item names, descriptions, prices. Correct anything it misread, then publish — the items land in the live menu, ready to sell on POS, QR ordering and delivery.

  • Photos, scans and multi-page PDFs, handled one branch at a time
  • Categories, items, descriptions and prices, kept in the order they were printed
  • A review screen before anything is published — nothing goes live unchecked
  • A second outlet’s menu becomes a five-minute job instead of an evening
AI Menu Upload — dinner-menu.pdf
Extracting page 3 of 3100%
Preview — Starters
Garlic Bread with Za’atar$4.50
Soup of the Day$5.00
Preview — Mains
Butter Chicken$12.90
Lamb Rogan Josh$14.50
Price unreadable — needs a lookReview
42 items · 6 categories Draft
Edit items
Publish to menu
Review FirstYou publish, not the model
Guest-facing AI

A chat assistant that has read your knowledge base

The customer chatbot answers from a knowledge base you write and control — opening hours, allergens, the delivery radius, today’s specials, your refund policy — instead of improvising. It holds the last dozen turns of a conversation, is rate-limited per visitor, and is instructed to decline politely and steer back when a question has nothing to do with your restaurant.

  • Answers grounded in chunks you authored, per restaurant and per branch
  • Injection-guarded: whatever a visitor types is treated as data, never as instructions
  • Hands off to your contact details rather than inventing an answer
Ask the restaurant
Do you have gluten-free mains?
Yes — nine mains are marked gluten-free tonight, including the Lamb Rogan Josh and the Grilled Sea Bass. The kitchen is not a gluten-free environment, so we flag cross-contact on request.
Do you deliver to Kingsford?
Kingsford is 4.2 km out, just inside our 5 km radius. Minimum order is $25 and the current window is 35–45 minutes.
Answered from knowledge base12 chunks
Book a table
How it works

Proposal, evidence, approval, audit

  1. 1
    Step 1

    Turn the module on

    Enable AI in the admin and supply the provider key for your installation. Until you do, nothing leaves your server and the tools sit dormant.

  2. 2
    Step 2

    It builds the context pack

    Orders, menu costs, stock, table state and rota for the branch and window in question are assembled server-side, scoped to the branches your account may see.

  3. 3
    Step 3

    It proposes, with the numbers

    Every recommendation arrives with the query results behind it, so you can disagree with it on the evidence rather than on a hunch.

  4. 4
    Step 4

    A human approves

    Auto-Pilot actions wait for you. If one is safe and boring enough to be routine, you can grant it standing permission — one specific action at a time, never a blanket switch.

  5. 5
    Step 5

    It logs, and it can be undone

    Executed actions are written to an audit trail with what they changed. Reversible ones carry an undo that restores the previous state exactly.

Honest limits

What this is, and what it is not

Assistive, not autonomous

The suite proposes and explains. A person decides. The only actions that can fire on their own are the ones you have explicitly marked safe, and those remain reversible.

It shows its working

Findings come with the figures behind them. Where the sample is thin — a new branch, a dish with nine orders all month — the tool says so instead of manufacturing confidence.

It is still a language model

Prices, costs and counts come from your database and are exact. The prose around them is generated, so read a campaign or a description before you publish it.

Who gets the most from it

Built for owners who do not have an analyst

Independent restaurants

One owner, no data team, and a Sunday afternoon to work out where last month went. The Brain and the Profit Leak radar do the query-writing you were never going to get to.

Multi-branch groups

Compare branches on the same context pack, spot the outlet discounting twice as hard as the rest, and roll out a price change once the Simulator has modelled it.

Delivery-led kitchens

Channel Positioning prices each marketplace separately, Menu Engineer keeps thin-margin sellers honest, and Tomorrow Plan sizes prep against the demand curve.

Straight answers

The questions owners actually ask

Model calls go to the AI provider configured for your installation. Customer personal data can be stripped from every prompt at a single choke point before it leaves, and prompts are built to carry only the figures a tool actually needs — order totals, item costs, stock counts — rather than raw customer records.

Get started today

Give your restaurant an analyst that never sleeps

The AI suite ships inside RestioX. Start a trial, load a menu with AI Menu Upload, run a few services, then ask the Brain what it noticed.

Start Free Trial