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.
18
AI tools inside the admin — no second app, no extra login
30-day
Rolling window of your own paid orders
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
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
Eight detectors run over your own figures: over-discounting, dishes sold below cost, drawers that do not balance, stock about to expire, refund spikes, popular items on a thin margin, supplier price rises and over-staffed shifts. 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 Brain
Ask questions about your own restaurant in plain language.
Auto-Pilot
Proposes concrete next actions; you approve, dismiss or undo.
Menu Engineer
Sorts dishes into stars, plowhorses, puzzles and dogs.
Dish Doctor
Diagnoses why one dish is underperforming and what to change.
Profit Leak
Finds and costs the six ways margin quietly disappears.
Live Ops
Real-time read on the service floor while service is running.
Kitchen Commander
Throughput and prep guidance for the pass during a rush.
Tomorrow Plan
Next-day covers, staffing and prep brief with a demand curve.
Photo Coach
Critiques a menu photo and says exactly how to reshoot it.
Taste Profile
Builds a taste model per returning guest from order history.
Personalized Menu
Reorders the menu a guest sees around what they actually order.
Table Mood
Scores every seated table 0–100 on how the visit is going.
Simulator
Model a price or staffing change before you commit to it.
Competitor Spy
Benchmarks your dishes against nearby competitors you add.
AI Insights Engine
Ask anything about your restaurant’s performance — “Ask RestioX” in the admin’s top bar opens it.
One-Click Campaign
Drafts a whole campaign — audience, offer, subject line, copy.
Restaurant Memory
Durable notes the AI writes to and reads back weeks later.
Voice Assistant
Ask 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.
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
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
How it works
Proposal, evidence, approval, audit
- 1Step 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.
- 2Step 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.
- 3Step 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.
- 4Step 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.
- 5Step 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
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.
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.
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
Menu Engineer keeps thin-margin sellers honest, and Tomorrow Plan sizes prep against the demand curve.
Straight answers
The questions owners actually ask
Does my restaurant data leave my server?
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.
Can Auto-Pilot change my menu or prices on its own?
Not by default. Every proposal waits for a human to approve or dismiss it. You can mark specific low-risk actions as safe to run automatically — hiding items that have been sold out for days, for example — and those stay reversible, with an undo that restores exactly what changed.
What happens when the AI gets something wrong?
You dismiss the proposal, or you undo the action. Nothing the suite produces is applied silently: every executed action is written to an audit trail with what it touched and when, and a proposal you have already acted on goes quiet for fourteen days instead of reappearing.
Will this run up an unpredictable AI bill?
Each restaurant has a monthly token allowance from its plan, and that allowance is checked before every single model call. The gate fails closed — when the allowance is spent, calls stop rather than keep charging — and an absolute ceiling sits behind it to catch runaway retry loops.
How much history does the AI need before it is useful?
Most tools read a rolling thirty days, and the forecasts compare the same weekday across earlier weeks. A brand-new branch will get thin answers for its first few weeks, and the tools say the sample is small rather than dressing a guess up as a finding.
Do I have to learn another interface?
No. Every tool lives inside the RestioX admin you already use. A Cmd+K palette opens any of them from any screen, the morning brief is waiting on the dashboard when you sign in, and the Voice Assistant takes spoken questions when your hands are busy.
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.