Proudly made in Europe

Audit the physical world from a single photo.

ASAI Labs builds computer-vision protocoling and auditing for FMCG operations - bakeries, gas stations, retail shelves and beyond. Your team snaps a photo; our custom vision models return verified, structured data in seconds.

Backed by

Trusted by FMCG operators across Europe

200+: Locations under management
0M+: Photos processed
3: Countries served

01/Where it works

One platform, every FMCG floor

The same vision pipeline audits any physical retail surface. We tune a custom model per vertical - bakeries are our flagship, with more rolling out across the field.

Flagship

Bakeries

Track fill ratios across display cases and baskets, flag sell-outs, and cut end-of-day waste - our most mature, custom-trained model.

Gas stations & forecourts

Audit forecourt stock, cooler facings, and convenience-shelf availability across every site from one dashboard.

Retail & grocery

Shelf availability, planogram compliance, and out-of-stock detection from a single staff photo.

Convenience stores

Real-time counter and cooler inventory without a single manual count.

Pharmacies

Verify shelf compliance and stock levels for regulated, fast-moving SKUs.

Your vertical next

If it lives on a shelf, counter, or forecourt, we can train a model for it. Tell us what you need to audit.

Flagship use case/Bakeries

Tired of empty display cases?
Stop letting missed sales cut into your margins.

Here is exactly how an ASAI Labs deployment works on a bakery floor.

02/The Process

How It Works

From a shop-floor photo to live dashboard data - in seconds.

Step 1

Snap a quick photo

Your staff receives a gentle notification to take a picture of the display cases. No complex hardware installations or clunky scanning wands are required - just a quick photo using any smartphone or tablet.

Employee taking a photo of a bakery counter

Step 2

Custom vision AI processes the image

The images are instantly routed to our custom-trained Machine Learning vision model. Designed specifically for bakeries, it effortlessly recognizes overlapping artisan loaves, delicate pastries, and varying tray layouts.

AI vision model identifying different types of bread

Step 3

Extract precise fill ratios

The system analyzes the visual data to calculate exact fill ratios for every shelf and basket. It knows immediately what is fully stocked, what is running low, and what has sold out, with clinical precision.

Visual overlay showing percentage of counter fill ratios

Step 4

Manage everything from a sleek dashboard

All extracted data is instantly aggregated into a centralized, minimalist dashboard. Monitor real-time availability, track sales velocity, and reduce end-of-day waste without ever doing a manual count.

Exponata dashboard showing live bakery inventory analytics

03/Performance

Live in weeks, accurate from day one

A custom model per vertical, trained on your real shelves - fast to deploy and precise once it is running.

2 weeksImplementation & trainingFrom your first batch of photos to a production-tuned model, live across every location - two weeks, max.

99.4%Recognition accuracyAcross products, display layouts, and real-world lighting conditions.

InstantPhoto to dashboardEvery capture is analysed, audited, and aggregated the moment it lands - no waiting.

$0Hardware & CapExNo clunky cameras, no sensors, no install. Just the phone already in your team's pocket.

Sneak peek/The platform

Every location, one screen.

Every location on one live map, colour-coded by fill ratio
Region, protocol and schedule management built in
Full CSV import / export - plug it into what you already run

app.asailabs.com / locations

Exponata platform showing locations across the country on a live map

04/Privacy

Faces blurred. Always.

Auditing the physical world means people walk into frame. They never make it into your data - every face is blurred before a photo is processed.

As-built survey blueprint of a retail aisle with the shopper's face redacted behind a privacy mask

Faces blurred on capture

Every photo passes through a face-detection and blur step before it is ever stored or analysed.

We read shelves, not people

Our models are trained on inventory geometry - fill ratios and product shapes - never on identities.

No biometric data, ever

We never extract, match, or retain facial features. There is nothing to leak.

Processed in the 🇪🇺

Built and hosted in Europe, GDPR-compliant by design.

RetailRetailrunsrunsononwhatwhatyouyoucan'tcan'tsee.see.ASAIASAILabsLabsmakesmakeseveryeveryshelf,shelf,countercounterandanddisplaydisplaycountable,countable,comparablecomparableandandaudit-ready.audit-ready.

05/FAQ

Everything you need to know

The essentials before you install.

Do I need to install expensive cameras?

Zero hardware required. Your team simply uses the smartphones or tablets they already own. No wiring, no installers, and no complex maintenance - just an app that works.

Are you recording my customers?

Never. Our AI is built with privacy-first architecture. It is trained to ignore faces and focus exclusively on "inventory geometry" - extracting fill ratios from your counters while keeping your customers anonymous.

Will this replace my employees?

Not at all. We believe bakers should be baking, not holding clipboards. Our system empowers your staff by removing the most tedious part of their job: manual inventory counting.

How does this differ from my POS data?

Your POS tells you what you sold; we tell you what is actually left on the shelf. We bridge the "visibility gap" to identify theft, waste, and restocking opportunities that sales data alone cannot see.

How accurate is the recognition?

Surgical. Our custom-trained vision models can distinguish between sourdough, rye, and baguettes even when they are overlapping or tightly packed in display baskets.

How fast can we see results?

Immediately. Once the first photo is snapped, our ML model processes the fill ratios in seconds, and your dashboard populates with actionable data instantly.

Can it handle low-light or crowded counters?

Yes. The system was designed for the "beautiful chaos" of a busy bakery morning. It compensates for varied lighting and recognizes products even through glass display cases.

What happens to the images after processing?

Security is paramount. Images are processed through an encrypted pipeline. Once the inventory data is extracted, you have full control over whether images are archived for audit trails or instantly deleted.