Skip to main content
Setting up a restaurant management platform usually means hours of manual data entry. AI Onboarding cuts that down dramatically. You bring the documents, Garde’s AI does the heavy lifting — and you stay in control the whole time.

​
What AI Onboarding Helps With

​
Recipes

Typing in every recipe ingredient by ingredient is nobody’s idea of a good time. AI Onboarding can:
  • Import recipes from documents you upload — PDFs, spreadsheets, Word docs, even photos of handwritten recipes.
  • Match ingredients to existing products in Garde automatically.
  • Suggest conversions when recipe units don’t match how you buy (ounces vs. pounds, each vs. case).
  • Flag items for review when it’s not confident in a match. You always get the final say.

​
Unit Conversions

Getting conversions right is the difference between accurate food costs and numbers that don’t mean much. AI Onboarding helps by:
  • Suggesting standard conversions (pounds to ounces, cases to each, and so on).
  • Learning from your corrections — every time you adjust a conversion, the system gets smarter.
  • Highlighting mismatches between purchasing units and recipe units so nothing slips through.

​
Vendor Configuration

Setting up vendors correctly means your invoices process accurately from day one. AI Onboarding assists by:
  • Auto-detecting vendor info from uploaded invoices.
  • Suggesting category mappings based on what’s on the invoice.
  • Matching vendor items to products so your data is connected right out of the gate.

​
Importing Inventory Items

AI Onboarding gives you two ways to import multiple items at once:

​
Upload a File

Navigate to the import screen and upload a file containing your items. Garde accepts CSV, XLSX, PDF, images, and TXT files. The AI parses the contents and extracts individual items for you to review.
The AI parser may cut off at approximately 100 items per upload. If your file contains more than that, upload the remainder in a separate batch.

​
Type Items Manually

Prefer to skip the file? Choose the Type option and enter one item per line, including the count unit. For example: 
Chicken Breast 1lb
Heavy Cream 1qt
Olive Oil 1gal

​
Reviewing Parsed Items

After uploading a file, you land on the Review Parsed Items page. Click through each item to confirm or edit the Name, Category, Count Size, and Default Price (optional). For each item, Garde’s AI agent suggests a category. Click “Approve & Next” to accept the suggestion, or type a new name to create or select a different category.
Watch for typos when creating categories. A small misspelling (e.g., “Prodce” instead of “Produce”) will create a duplicate category. If this happens, go to Items, select the incorrect category, click the item, edit it, and change it to the correct category. To move all items out of a wrong category at once, delete the category, select the new destination category, then confirm the delete and save.

​
Reviewing Parsed Recipes

After you upload recipe documents (PDFs, spreadsheets, images, or other supported formats), Garde’s AI parses them and drops you into the Review Parsed Recipes screen. The page title tells you how many recipes were extracted — for example, “27 recipes parsed from upload.” The screen is split into two panels:
  • Left panel — A scrollable list of every parsed recipe. Each recipe displays a status icon: an empty circle for recipes you haven’t reviewed yet, a green checkmark for recipes you’ve approved, and an arrow indicator for the recipe you’re currently viewing.
  • Right panel — The full detail view for the selected recipe, showing its position in the list (e.g., “Recipe 16 of 27”).

​
Reviewing Each Recipe

For every recipe, verify and adjust the following fields:
  1. Name — Confirm or edit the recipe name extracted by AI.
  2. Goal Cost % — Set your target food cost percentage for the recipe.
  3. POS Menu Items & Modifiers — Map the recipe to one or more POS items so Garde can automate stock depletion when those items are sold. The dropdown shows how many items are currently selected (e.g., “2 selected”).
  4. Sub-Recipe / Prepared Item — Toggle this on if the recipe is a prep item (sauce, dressing, stock, etc.) that will be used as an ingredient in other recipes.

​
Verifying Ingredients

Each recipe includes an ingredients table with the following columns:
ColumnDescription
Parsed NameThe ingredient name as extracted by AI from your uploaded document.
Matched ItemA dropdown to match the parsed ingredient to an existing inventory item. You can search for items or create new ones directly from this dropdown.
QtyThe quantity required. Editable.
UnitThe unit of measure (e.g., oz, slice, Unit). Select from the dropdown.
Yield %The yield percentage for the ingredient. Defaults to 100 — adjust downward to account for trim, peel, or other waste.
A warning triangle (⚠) next to a Matched Item means the AI is not fully confident in its match — for example, “Carrot Daikon” matched to “Carrot.” Always verify flagged matches and correct them if needed.
You don’t need to review every recipe in a single session. Green checkmarks in the left panel let you track your progress and pick up where you left off.

​
How to Get Started

  1. Log in to Garde and head to the Setup or Onboarding section.
  2. Follow the prompts to upload your recipes, confirm vendor details, and review suggested conversions.
  3. Review and approve the AI suggestions. Nothing gets finalized without your sign-off.

​
Tips for a Smooth Setup

  • Upload whatever you have. Polished spreadsheet? Great. Stained recipe binder photo? Also great. AI Onboarding works with messy real-world documents.
  • Correct early, benefit later. The more fixes you make during initial setup, the more accurate everything becomes going forward.
  • Don’t wait for perfect. You can refine recipes and conversions anytime. Getting your data in quickly means you start seeing value sooner — and your team can start using Garde right away.
Running into something weird during setup? Our onboarding team has seen it all. Reach out at contact@garde.app and they’ll get you sorted out.