How to use Promenu AI tools for your restaurant menu

In the Promenu panel, AI tools helps you raise a catalogue faster: parse a paper menu, fill dish cards, make photos and a cover, translate copy, and find gaps. AI never publishes for you — only what you apply or save reaches the guest menu.

Below are all six tools and a sensible order if the venue already exists. No place yet? Build the digital menu first.

What you need

• Access to the Promenu panel and a selected venue.
• Either a few items already in the menu, or photos, a PDF, or an online-menu URL.
• A clear head that composition and nutrition from AI are an estimate, not a lab result.

Where to click in the panel

1. In the sidebar open AI tools — that is the label outside the Russian locale.
2. Pick the venue if you have several: the tools run in the context of the selected place.
3. Jobs sit on the left, the workspace on the right. Monthly quotas show in the section itself: how many are left this month, and how many the next run will use.

Which AI tools Promenu has

Where to start: a useful order

You do not have to run everything on day one. Import, photos on a few bestsellers, and a cover are often enough to go live.

Menu from sources

Why: skip typing the catalogue by hand when the menu already exists. Path: AI tools → Menu from sources.

1. Add a source: Menu photos or PDF, up to 5 files — or an Online menu URL. You need the page that lists dishes, not the whole marketing homepage. You can attach screenshots next to a URL.
2. Set the Menu language and extra instructions if you need them.
3. Press Parse menu. You can leave the tab — the result is saved and appears here when it is ready.
4. Proof the draft: names, prices, categories. Categories the place already has are reused.
5. Only after that review press Apply to menu — nothing reaches the catalogue before that click.

Menu analysis

Why: see which cards are hollow before you invite guests. Open it from AI tools → Menu analysis or from Menu → Items with the Menu analysis button (the tooltip says "Menu completeness analysis").

1. The tool scans the selected venue and shows a readiness percent.
2. Filter the gaps: no photo, description, composition, nutrition, weight, price, or category.
3. Expand a row → Fix and edit fields in place, or open the full product card.
4. Save the edits. The remaining holes are closed by Fill dish details, photos, and translation.

Analysis is a check, not a generation: there is usually no separate quota to run it.

Fill dish details

Why: get card copy fast when you have a name and, ideally, a photo. Path: Fill dish details.

1. Select dishes. Filters: No description / No composition / No nutrition, category, search, "Only ready to generate".
2. Under Generate mark the fields: description, composition, nutrition — you do not have to take all three.
3. Press Generate. Before the run you see how much quota it will use.
4. Review the suggestions and press Apply selected — or apply description, composition, and nutrition one by one.

Menu photo generation

Why: drop the grey placeholders — enhance a phone shot or draw the dish if there is no photo yet. Path: Menu photo generation. The same flow opens from a product card → Images.

1. Pick a Processing style — from "Minimal changes — real photo" to a studio, warm light, or a dark frame. One style on every hit looks more coherent than ten different universes.
2. Optionally set Backdrop and notes: a backdrop color (linen, cream, wood, marble, custom) and a short note — up to 200 characters. You can skip this; the style picks the surface.
3. Select one or more dishes.
4. Press Start generation. The panel’s own hint is about 1–2 minutes per photo; keep the page open while results arrive.
5. Compare Before / After and press Apply selected. The old photo stays until you confirm the replacement.

Venue cover

Why: a venue card in the app without hiring a designer. Path: Venue cover or Places → the place → Profile → Images.

1. Check "What AI will use": name, description, cuisine types, categories, and dishes.
2. Optionally upload a photo of the room, facade, storefront, bar, sign, or logo. Or press Without my photo — the cover is built from place data.
3. Write image wishes — light, angle, mood, where to put the logo. Up to 500 characters. Do not ask for people: AI removes them.
4. Press Generate cover. Usually 1–2 minutes; you can leave the tab — the result is saved.
5. If the frame works — Apply to gallery or Set as main photo. The first photo in the gallery is often the cover the guest sees.

Smart translate

Why: guests from other countries see names, descriptions, and composition in their language. Path: Smart translate. It fills Name, Description, and Composition. The source language is the venue language.

1. Open the tool — it finds empty and outdated translations. Up-to-date texts are left alone.
2. Check the target menu languages. The set depends on the place: English, Español, Português, ไทย, 中文, Қазақша, 日本語, and any others you enabled.
3. Press Start translation. It usually takes a few minutes; you can leave the tab.
4. Refresh scans missing fields again.
5. Open the guest menu in that locale and fix 5–10 flagship dishes if a line sounds foreign.

How not to burn quota for nothing

• Do not run AI on the whole menu the first evening. Start with the 5–10 dishes you actually sell.
• Before a run, read "will use …" and the remainder under "Quotas this month".
• Import → proof → Apply to menu. Generating blind creates more work later.
• One photo style on the hits beats ten different treatments in one catalogue.

Checklist before you invite guests

• I know where AI tools lives.
• If I needed it, I built a draft in Menu from sources and pressed Apply to menu only after a review.
• I ran Menu analysis and closed the critical gaps.
• Bestsellers have a description and a photo.
• The venue cover is in place.
• If other languages are needed — I ran Smart translate and proofed the flagship dishes.
• I did not burn the quota on the whole menu at once.

Where to click if you forgot

Common questions