Pilot Program · Now Open

Inventory that thinks ahead.

The inventory platform for modern food service. We replace gut-feel ordering with a data-driven system that learns your operation — so you stop losing money to waste and stockouts.

Request a demo +32 468 53 34 97
GDPR compliant
Tenant-isolated
EU-hosted
aresai.ai / dashboard / overview
Revenue · 7d
€14,284
↑ 8.2%
Margin
68.4%
↑ 1.1pts
Waste Cost
€312
↓ 24%
Forecast · p90
94.2%
acc
Demand Forecast · Next 14 Days
p10 · p50 · p90
Reorder · Tomorrow 08:00
NEEDS APPROVAL
Flour 0025 kgMilan Mlin€21.50
Whole Milk40 LDobra Dolina€38.00
Hazelnut Spread6 kgDelmar€54.90
3 suppliers · delivery Tue 07:00€114.40
Built for
RestaurantsCafésBakeriesCateringGhost kitchensMulti-location chains
01The Industry Reality

Food service margins are brutal.
Every inefficiency compounds.

4–10%
of food purchased becomes waste before it's ever served
28–35%
of revenue goes to food costs — the single biggest variable expense
3–5%
average net profit margin in the restaurant industry
Sources: USDA Economic Research Service · Fourth / NRDC Food Waste Study · National Restaurant Association 2025 · Toast 2025
The operators who survive the next decade of food service won't be the ones with the best marketing. They'll be the ones who know — precisely — how much to order, how much to prep, and what's about to expire.
02What We Solve

Six problems costing you money right now.

Over-Ordering & Waste
You order on gut feeling. Food expires. Money goes into the bin — week after week.
Stockouts & Lost Sales
Run out mid-service. Customers leave. A single popular dish going 86 costs you the full ticket.
Hours on Manual Ordering
Your manager spends 5–10 hours a week counting stock and calling suppliers. That's labor, not strategy.
No Visibility Ahead
You react after problems happen. No forecasting, no early warning, no way to plan for next weekend.
Prep Guesswork
The kitchen over-prepares to avoid running out. Excess gets discarded at close every single day.
Expiry Surprises
Ingredients expire in the walk-in. Discovered mid-service. Scramble for a substitute. Repeat.
03How It Works

A closed loop, running every night.

ARES reads your POS, forecasts demand per product, explodes that through your recipes to the ingredient level, and builds optimized purchase orders per supplier — all before you open in the morning.

Step / 01
Predict
LightGBM quantile regression models forecast demand for every menu item, 14 days ahead. Weather, day-of-week, events, and trends all factored in.
Step / 02
Order
The engine converts product forecasts into ingredient needs via your recipes, then builds cost-optimized purchase orders per supplier. One-click approval.
Step / 03
Prep
Every morning at 5:30 AM, the kitchen receives a prep plan with exact batch quantities based on today's forecasted demand. No more over-prepping.
Step / 04
Track
Real-time inventory updates from every sale. FIFO batch tracking. Expiry alerts. Complete audit trail of every gram in and out.
04Platform

Built like enterprise software.
Priced for independent operators.

ML
Quantile Demand Forecasting
Per-product LightGBM models with multi-quantile outputs (p10–p98) so you can tune reorder aggressiveness per ingredient.
Optimizer
Supplier-Aware Reorder Engine
Optimizes across lead time, MOQ, case size, delivery fees, tiered pricing, cutoff times, and per-ingredient shelf life.
Ops
Morning Prep Planning
Computes optimal small / medium / large batch mix to minimize daily discards, delivered by email at 5:30 AM.
Data
FIFO Batch Tracking
Every incoming delivery creates a traceable batch. Usage allocated FEFO for accurate cost-of-goods and expiry visibility.
Real-time
15-Minute POS Sync
Orders flow from your POS into ingredient-level usage every 15 minutes via deterministic, idempotent ingestion.
Self-heal
Modifier Rate Learning
The system compares recipe-predicted vs actual ingredient usage and auto-corrects for systematic over and under-pour.
05Security & Data

Your operation is yours.

We handle sales data, recipes, and supplier relationships — the sensitive core of your business. We designed the platform assuming everything in it is confidential.

Multi-tenant isolation at the row level
Every query is scoped by organization and location. A client's data is inaccessible to any other tenant — enforced at the database layer, not just application code.
Encrypted at rest and in transit
TLS 1.3 for all API traffic. AES-256 encryption for data at rest. Credentials handled through industry-standard JWT tokens with scoped permissions.
EU-hosted infrastructure
Data stored in EU data centers (Frankfurt). No data leaves the European Union. GDPR-compliant data handling from day one.
Role-based access control
Owner, manager, staff, and viewer roles. Destructive operations require manager approval. Complete audit trail of every stock-affecting action.
06Frequently Asked

Questions operators actually ask us.

Accuracy depends on your data volume and how stable your operation is. We publish forecasts as quantile ranges (p10 to p98) rather than a single number, so you can tune how aggressive you want to be per ingredient — conservative for perishables, aggressive for shelf-stable items. During onboarding we typically need 60 to 90 days of POS data to reach meaningful accuracy.
The simplest way to get started is our CSV import workflow — you export your sales history, we load it, the forecasts start running within 24 hours. For long-term, we work with you on direct POS integration. Tell us which POS you use and we'll walk you through the options.
It stays yours. We never sell, share, or use your data to train models for other clients — each restaurant's models are trained only on that restaurant's own data. You can export everything at any time, and we delete your data on request if you leave.
Honest answer: the AI will be wrong sometimes — any forecast is wrong sometimes. That's why we use quantile forecasts, not point estimates. You control the safety buffer per ingredient. The system also tracks its own accuracy and surfaces it in your dashboard, so you always know how much to trust it on any given day.
The first usable forecast appears 24 hours after we have your menu, recipes, suppliers, and at least 60 days of POS history loaded. A realistic timeline from first call to operational system is 1 to 2 weeks, most of which is collecting and cleaning your existing data.
Pricing is per location, scaled to your revenue. We don't publish a number because the right price depends on whether you're a single cafe or a multi-site operation. Book a call and we'll give you a straight answer within the first 10 minutes.
No. ARES runs alongside your existing operation. Your team keeps working the way they already work — the difference is they're now working from a daily prep plan and pre-approved reorder list instead of guessing. Most operators see it as less change, not more.
— Ready when you are —

Stop guessing. Start knowing.

20 minutes. We show you the platform running against real forecast data. You decide if it's worth another conversation.

Book a demo +32 468 53 34 97
07Get in Touch

Tell us about your operation.

A 20-minute demo against real forecast data. No pitch deck, no theatrics — just the platform running.

Or call directly: +32 468 53 34 97