A catering order by text, in the customer’s language
The problem
Catering requests arrive by text during the lunch rush, often in Spanish. They wait for someone to answer, and pickup windows get promised past what the kitchen can pack.
How it's resolved
The AI reads each text in the customer’s language, with an English translation for the staff, and writes every reply. Code prices the order from the menu, checks the pickup window she asked for against its capacity and offers the nearest windows with room; the books themselves refuse an oversell. An allergy is the exception: code matches it to the dishes that hold it, and only the owner answers, in her own words, which the AI translates. Code then books the places, drafts the order and sends a link to the caterer’s own deposit page, and the customer pays on her phone.
What you'll see
- 01A text in Spanish: tacos for 60, eight of them vegan, next Friday at 11:30
- 02The order builds on the counter tablet in English, priced from the menu: $982.00
- 0311:30 has 40 places left, so the order runs red past its line, and the books refuse the 60 places themselves
- 04Code offers noon and 12:30, and the AI writes the reply in Spanish, saying it is an automated assistant
- 05She takes noon and mentions a sesame allergy: only the owner answers, and the AI translates her words
- 06Code books the 60 places at noon, and she pays the $245.50 deposit on the caterer’s own page
Under the hood
Each business term in the video, and the Traide building block it runs on.
| In the business | On Traide |
|---|---|
| Pickup window | Capacity pool, with a start and an end |
| A guest’s place | Ticket variant drawing on the window’s pool |
| An oversell | Refused at checkout: insufficient stock |
| Menu and allergens | Catalog, with allergen tags in product metadata |
| Catering order | Draft order, sent to the customer |
| Deposit | Partial payment on the order |
Guardrail. Only the owner answers about allergens: a vegan request is not an allergen promise. The first reply says it comes from an automated assistant.
- Reading and writing in the customer’s language
- Translation for staff
- Grounded drafting
- Capacity checked in code
Have a problem like this one?
You'll talk to a senior engineer, not a sales rep, and leave with a clear first move — whether or not we work together.