Graphics to Text — Fields Builder
Graphics to Text is a free AI tool that extracts menu content from images and PDFs and converts it into a structured CSV file ready to import into Menuboard Manager. Got a menu on paper, buried in a photograph, or locked inside a multi-page PDF? Upload it and get a clean table of item names, descriptions, and prices back in seconds — no manual retyping required.
The tool is available to anyone at no cost. A quick one-time registration (name and email) is required before your first extraction.
What It Does
Graphics to Text solves the content migration problem that almost every new Menuboard Manager customer faces: your menu already exists — in a printed menu, an old PDF, a photo of your chalkboard — but getting that content into Menuboard Manager fields has historically meant retyping everything by hand.
Graphics to Text reads the visual content of your uploaded file using AI vision and returns a structured table of:
Section headings (labeled as type
text)Menu items with name, description, and price (labeled as type
card)Size and variant rows for items with multiple sizes or add-ons (e.g.,
Latte - Small,Latte - Large)
The result downloads as a CSV file pre-formatted for direct import into Menuboard Manager as a fields data source.
How to Access
Go to menumagik.ai/graphics-to-text, or navigate there from the Menu Magik home page by clicking the Graphics to Text tool card.
The tool is also linked in the footer of the Toaster and Magik Menu Builder apps.
Step-by-Step Workflow
Step 1: Register (First Use Only)
On your first extraction, a registration modal appears asking for your first name, last name, and work email. This is a one-time step — your credentials are stored locally in your browser for 30 days. You will not be asked to register again on the same device during that period.
Note: Registration is free and does not require a credit card. You will be opted in to Menuboard Manager product updates by default; uncheck the consent box if you prefer not to receive them.
Step 2: Upload Your Menu File
Drop your file onto the upload zone or click it to browse. Accepted formats are:
Format | Notes |
JPEG / JPG | Photos, scans, screenshots |
PNG | Screenshots, exported graphics |
Single or multi-page menus |
After selecting a file, a preview appears — an inline image thumbnail for photos, or an embedded PDF viewer for PDFs. A file name and size confirmation shows below the drop zone.
Tip: The clearer the source material, the better the extraction. High-contrast, well-lit photos work best. Avoid images where text is obscured by overlapping graphics, watermarks, or decorative backgrounds.
Step 3: Extract Menu Data
Click Extract Menu Data. The tool sends your file to the AI for analysis.
Images are processed as a single page and return one result set.
PDFs are processed page by page. Each page produces its own named tab in the results and its own CSV file on download.
Extraction typically takes 10–30 seconds depending on file size and complexity. A spinner and status message appear while the request is in flight.
Step 4: Review and Edit
When extraction completes, the Review & Download section appears with a table showing all extracted rows. For multi-page PDFs, page tabs appear above the table — click any tab to switch between pages.
Each row in the table contains four editable fields:
Column | What it contains |
Name | Item name or section heading |
Description | Item description (blank for headings) |
Price | Numeric price, stripped of currency symbols |
Type |
|
You can edit any cell inline by clicking it. Additional controls:
+ Add Row — appends a blank row at the bottom of the current page's table
✕ (delete row) — removes a row from the current page
Type dropdown — change a row's type; rows marked
textdisplay with a subtle highlight to distinguish headings from items
The row count for the active page is shown in the lower right of the table.
Tip: Always do a quick pass over the results before downloading. AI extraction is accurate for most menus, but dense layouts, handwritten text, or decorative fonts may produce occasional errors or missed items.
Step 5: Download the CSV
Two download options are available at the bottom of the results section:
Download Page CSV — downloads the CSV for the currently selected page tab
Download All CSVs — downloads one CSV file per page in sequence (with a 200ms stagger between files to avoid browser download conflicts)
Downloaded files are named mbm-[page-name].csv — for example, mbm-drinks.csv or mbm-page-1.csv.
Output Format
Each downloaded CSV uses the standard Menuboard Manager fields import format with the following columns:
Id, Name, Value, NotAvailable, CardName, CardDesc, CardPrice, CardIconName, CardIconUrl, CardIconThumbnailUrl, CardIconMimetype, Placeholder, Type, ShowOnPricesheet, Classes, PosId, NoLeadingZero, NoLeadingDecimal, NoCents
For card type rows, the CardName, CardDesc, and CardPrice columns are populated from the extracted data. For text type rows (section headings), only the Name column is set. ShowOnPricesheet defaults to TRUE for all rows.
Currency symbols are stripped from prices during extraction — prices are stored as digits and a decimal point only (e.g., 9.99, not $9.99).
Note: The Id and PosId columns are left blank in the downloaded CSV. If you are connecting fields to a POS integration, use Toaster to map and populate those values after import.
Tips for Best Results
Use the highest-quality source you have. A PDF exported from design software will extract more accurately than a phone photo of a printed menu.
Crop out non-menu content before uploading. The AI ignores restaurant name, address, phone, website, hours, social handles, allergen disclaimers, and promotional text — but removing these from the image or PDF entirely reduces the chance of any noise appearing in results.
Multi-page PDFs work well. Each page is processed independently and gets its own named tab. If your PDF has a logical page structure (e.g., "Lunch", "Dinner", "Drinks"), those names are often used automatically for the tab labels.
Items with sizes produce variant rows. For an item like "Latte" with Small and Large sizes, the extractor creates two rows:
Latte - SmallandLatte - Large. If you want to consolidate these, delete the variant rows and edit the base item manually.Add missing rows manually. If the extractor misses an item — common with very small print or stylized fonts — use + Add Row to add it without re-running the extraction.
Limitations
Accepted file types: JPEG, PNG, and PDF only. Word documents, Excel files, and other formats are not supported.
Single image = one page. Images always produce a single-page result regardless of how much content they contain.
Maximum token output: The AI response is capped at 4,096 tokens. Very dense menus with hundreds of items may be truncated. For large menus, consider splitting the PDF into sections before uploading.
Handwriting: Handwritten menus may extract with lower accuracy than printed or typeset menus.
Non-English text: The extractor supports multilingual menus, but accuracy may vary for non-Latin scripts.
No image upload from within Menuboard Manager: Graphics to Text is a standalone web tool at
menumagik.ai/graphics-to-text. It is not embedded inside the Menuboard Manager dashboard.Registration required: First-time use requires a name and email. The token is stored in your browser's local storage and expires after 30 days.
Related Articles
Toaster — Toast Field Builder — After importing your Graphics to Text CSV, use Toaster to map POS IDs to your Menuboard Manager fields for live price and availability syncing.
Magik Menu Builder — Use an AI-generated menu board layout as your design starting point instead of building from scratch.
Working with Presentations — Learn how to add your imported fields to a presentation and configure dynamic content on your menu board.