The Margin Leak You Can’t See: MenuMargin AI Turns Supplier Invoices into Better Menu Prices

Vansh Dixit — Sun, 18 Oct 2026 17:48:38 +0000

What I Built
Restaurant owners usually discover margin problems after the money is already gone.

MenuMargin AI is a real-time food-costing and inflation-protection dashboard for restaurants, bakeries, cafés, food trucks, and other small food businesses.

Demo link:
https://menumargin-ai.onrender.com/

A business owner can:

Create a free business workspace.
Add ingredients, packaging, labor, recipes, selling prices, portions, and monthly sales volume.
Upload or paste supplier invoices.
Convert commercial case quantities into usable culinary units such as cost per ounce, egg, gram, or fluid ounce.
Track wholesale inflation across the pantry.
Calculate plate cost, food-cost percentage, gross margin, and monthly margin leakage.
Identify dishes that are above their target food-cost threshold.
Ask Gemma for recommended actions such as:
Raising a menu price.
Switching suppliers.
Adjusting portions.
Rebalancing recipe ingredients.
Apply recommendations back to the business workspace.
Generate a one-page profit audit.
The product is designed for owners who need an answer in three seconds:

Am I losing money, where am I losing it, and what should I do next?

The people most likely to pay for this are independent restaurant owners, bakery owners, café operators, food-truck owners, and hospitality consultants who cannot justify expensive enterprise food-costing software.

The demo includes seeded bakery data so you can immediately explore:

Dashboard and margin alerts.
Invoice Scanner.
Recipe Studio and COGS.
Inflation Radar.
One-page profit audit.
Gemma optimization suggestions.
First-user traction tracking.
Code
The project is built as a CommonJS Node.js and Express application with a vanilla HTML, CSS, and JavaScript frontend.

The main areas are:

The application includes:

Authenticated business workspaces.
Session cookies and bearer-token support.
Password hashing with Node.js crypto.scrypt.
SQLite-backed users, sessions, workspaces, and memberships.
Workspace-level authorization.
Isolated business inventory and recipe data.
CSV inventory import.
Invoice extraction and pantry updates.
Recipe recalculation after supplier-price changes.
Gemma-powered optimization workflows.
Responsive and reduced-motion-friendly UI behavior.
A public GitHub repository and deployed demo still need to be added before publishing this submission publicly.

How I Built It
MenuMargin AI is built around Gemma as the reasoning layer for culinary cost analysis.

The invoice workflow accepts supplier invoice text or uploaded files, then:

Extracts vendor and invoice metadata.
Identifies product lines and quantities.
Interprets commercial packaging such as cases, pounds, quarts, dozens, and cartons.
Converts those packages into culinary preparation units.
Calculates normalized unit cost.
Compares the new price with the existing pantry price.
Flags meaningful inflation changes.
Recalculates recipe plate costs.
Sends the resulting business context to the Gemma optimization workflow.
The frontend intentionally remains framework-free:

Vanilla HTML, CSS, and JavaScript.
Express backend.
Native CSS motion and requestAnimationFrame.
IntersectionObserver for progressive reveal effects.
Inline SVG visualizations.
Responsive layouts without forcing a framework migration.
The interface includes:

Dashboard KPI hierarchy.
Margin leakage hero card.
Food-cost target gauge.
Inflation sparklines.
Invoice scan-line feedback.
Staggered extraction results.
Gemma optimizer drawer.
Command palette.
Keyboard navigation.
Reduced-motion support.
Accessible live-region announcements.
Responsive mobile navigation.
Why Does Open Innovation Matter?
Food businesses deal with sensitive supplier information, negotiated prices, recipes, and margins.

A closed AI API can make owners uncomfortable because their invoices and commercial data may leave their environment. It can also introduce:

Vendor lock-in.
Unpredictable pricing.
Usage limits.
Data-retention concerns.
Less control over how recommendations are generated.
Open-weight AI makes a different model possible.

**With Gemma, MenuMargin AI can move toward a workflow where businesses can run inference locally or through an infrastructure provider they control. That means an independent bakery could eventually analyze invoices and recipes without sending confidential supplier data to a closed third-party system.

Open innovation also makes the product more adaptable. Culinary unit conversion, margin rules, supplier mappings, and optimization prompts can be inspected, tested, customized, and improved by the community.**

The goal is not just to make AI answer a question. The goal is to make the entire business decision system more transparent, affordable, and controllable.

My First User
No paid customer has been acquired yet.

The prototype is designed around the workflow of an independent bakery or restaurant owner who wants to understand supplier inflation and menu profitability without paying for enterprise software.

The most important validation question is:

Would a small food business pay a predictable monthly price to automatically detect margin leakage and receive actionable menu recommendations?

The current product direction is focused on proving that value before introducing paid plans.

DEV Community: Vansh Dixit

The Margin Leak You Can’t See: MenuMargin AI Turns Supplier Invoices into Better Menu Prices