Your AI team, and why it has many mouths
The platform doesn’t give you one chatbot. It gives you a team of specialists who share your venue’s live data, and three microphones through which that team speaks — one for guests, one for your staff, one for you. Same brain, different mouths, each tuned for who’s listening.
This page is for owners who wonder why there are many named agents instead of one, and why a guest, a waiter, and an owner get three different answers to the same question.
The team
Each specialist exists because a different problem needs a different kind of judgement:
| Who | What they fix |
|---|---|
| Marco — coordinator | You ask one question; he routes it to the right specialist and hands back one answer. You never need to know who does what. |
| Nina & Davide — market & competitors | The street you can’t watch yourself: competitor menus, prices, offers, review themes — read nightly, flagged with a drafted counter-move. |
| Elena — copy & brand voice | The posts and review replies you never have time to write, drafted in your voice. She never publishes alone. |
| Dani — style designer | Your phone photos, restyled into menu-ready shots that match the look you described once. |
| Bruno — chef & food expert | Dish ideas and seasonal angles, grounded in your actual pantry — not generic recipes. |
| Lucia — performance analyst | What’s working, in plain numbers: what guests open, where they linger, what’s climbing. |
| Giulia — offer strategist | What to feature, bundle, or push — your menu as a selling tool, priced against the street. |
They all read from the same rooms (Tell it once), so they never contradict each other — the copywriter describes the dish the chef’s recipe defines, priced as the till prices it.
The three microphones
The team also answers questions live, on three surfaces — and each surface has different rules about what it may say and do:
| The mouth | Who it serves | What it can do | What it never does |
|---|---|---|---|
| The guest assistant | A diner on your public menu | Recommend, explain, pair, escalate allergen questions to a human | Quote your costs, reveal internal notes, confirm allergen safety alone |
| The staff coach | A waiter at the till, mid-service | Everything above, plus internal notes, upsell logic, cross-contamination flags | Move money, override the kitchen |
| The owner chat | You, anywhere — in the app or on your phone | Everything above, plus act: change a price, hide a dish, pull sales numbers | Delete data or touch anything outside its logged tool set |
Each row sees strictly more than the one above, and that boundary is enforced in code — the guest mouth cannot leak the staff notebook.
The rule: each mouth is tuned for the cost of saying the wrong thing. A guest hearing a wrong allergen answer is a hospital visit, so the guest mouth escalates. A waiter can double-check with the chef, so the staff mouth explains how. You just want it done, so the owner mouth acts — and logs every action.
Worked example
“Is the pasta al limone gluten-free?” — three mouths, three correct answers. The guest hears: it contains gluten, a swap is possible, the kitchen will confirm cross-contamination. The waiter additionally sees the chef’s internal note that the boiling water is shared, and is told to switch to the allergen-pan setup. The owner, asking at midnight why it isn’t selling, gets the sales trend and the price change that caused it — and can act on the spot.
Related features
- The problems we fix — which specialist fixes what
- Centrale — your marketing team — where you meet the team
- How the AI thinks about your menu — the notebooks every mouth reads