aiR intelligence
Anomaly Detection
Void spikes, sales drops, cash anomalies — statistical detection against each branch's own baseline, with instant alerts.
How does anomaly detection work?
aiR intelligence builds a 'normal' profile (baseline) from each branch's historical data, then compares live data against that profile using the z-score method. There is no fixed threshold — what is detected is deviation from each branch's own normal.
How Does It Work?
From your POS data to anomaly detection in 4 steps
- 01
Baseline Learning
A normal behaviour profile is built from each branch's history, learning metrics such as void rate, average sales and cash percentage.
- 02
Z-Score Detection
Live data is compared with the baseline, and values that deviate markedly from the standard deviation are flagged as anomalies.
- 03
air AI Explanation
The detected anomaly is analysed by air AI and turned into a plain-English explanation the manager can act on.
- 04
Mobile Notification
The explanation and recommendations reach the branch manager instantly through the mobile app, and the event is recorded.
Types of Anomaly Detected
aiR intelligence continuously monitors these anomaly types
Void Spike
A sudden rise in the void rate is detected. It can point to theft, manipulation or a training gap.
Sales Drop
Sudden falls against expected sales volume are detected. They can signal an operational problem, understaffing or a technical fault.
Cash Anomaly
Unusual shifts in the share of cash payments are detected, and the risk of fraud or a till shortfall is assessed.
Excessive Discounting
A clear rise in discount rates against the norm is detected, pointing to a loss of control or unauthorised discounting.
A Real Scenario
The full story of one anomaly detection, start to finish
- Normal
Void rate: 0.5%
The Kadıköy branch's average void rate over the last 30 days is calculated at 0.5%.
- Detection
Void rate: 2.2% (z-score: 3.4)
Around lunchtime today the void rate jumped to 2.2%. A z-score of 3.4 means it sits 3.4 standard deviations away from normal.
- Analysis
air AI produces an explanation
"The void rate at your Kadıköy branch has risen 340% in the last three hours. Nine of the 12 voids were done by the same cashier. That cashier's training should be checked."
- Notification
Mobile notification sent
A mobile notification went instantly to branch manager Mehmet. The event was written to the ai_events table.
- Resolution
Issue resolved
The manager went to check. It turned out the new cashier was using the void process incorrectly. After training, the void rate returned to normal.
Our Approach
Statistical detection instead of fixed-threshold checks
Intelligent Detection with aiR intelligence
- Z-score: detection against each branch's own normal
- A baseline profile per branch
- A low false alarm rate
- A plain-English explanation from air AI
- Instant notification through the mobile app
Frequently Asked Questions
How does anomaly detection work?
aiR intelligence builds a 'normal' profile (baseline) from each branch's historical data, then compares live data against that profile using the z-score method. There is no fixed threshold — what is detected is deviation from each branch's own normal.
What kinds of anomaly are detected?
Void spikes (possible theft or manipulation), sudden sales drops (operational trouble), cash ratio anomalies (a fraud signal), excessive discounting (loss of control) and unexpected shifts in payment method.
What happens when an anomaly is detected?
When an anomaly is detected, air AI produces an explanation in plain English and the branch manager is notified instantly through the mobile app. The event is written to the ai_events table and the feedback loop keeps the system learning.
What is the false alarm rate?
A z-score based statistical method produces far fewer false alarms than a simple threshold check. Because each branch's own standard deviation is calculated, a mall branch and a high street branch are judged by different yardsticks.
What does setup require?
No extra installation. Data is pulled from your robotPOS system and analysed automatically. After a minimum of two weeks of accumulated data the baseline is built and anomaly detection goes live.
5 Intelligence Layers, One Platform
Each is fed by the same data pipeline, works together and reinforces the others.
Anomaly Detection
Void spikes, sales drops, cash anomalies — statistical z-score detection against each branch's own normal, with an instant mobile alert before the problem grows.
Explore in detailDemand Forecasting
Hourly, branch-level forecasts driven by weather, Ramadan, holidays and location type — data-led planning for raw material orders, shift rotas and stock levels.
Explore in detailMenu Intelligence
Product lifecycle (rising→declining), combo penetration, cross-sell opportunities and price elasticity. Understand why each product is climbing or falling.
Explore in detailIQ Assistant
"My three worst branches yesterday?" — ask in plain English and get an instant answer. A personalised AI briefing every morning at 06:00. The data comes to you; you do not have to go looking.
Explore in detailBranch Analytics
Five-dimension scoring (turnover, trend, efficiency, data quality, security), location-type benchmarking and an hourly heatmap. Compare mall with mall and high street with high street.
Explore in detail
Catch Anomalies with AI
Try aiR intelligence anomaly detection in your branch. No installation required — it starts working straight from your POS data.
Using a different POS system?
No problem — you do not need to replace your POS.
With Restaurant Intelligence as a Service, send your sales data every 3 minutes and get all of aiR intelligence's anomaly detection, demand forecasting, menu intelligence and branch analysis. RIaaS is made for you.
Why robotPOS?
The Easiest, Fastest and Most Reliable Restaurant Automation and Management System
Easy and Practical to Use
Solutions that adapt to your workflow and way of working — practical, quick to learn and dependable for years.
23 Years of Experience
Since 2003 we have built end-to-end management solutions, from boutique venues to restaurant and cafe chains with hundreds of branches.
Specialised Solutions
We work solely on restaurant and cafe automation, and we deepen that expertise every day.
Foresight with AI
aiR intelligence turns your sales data into foresight and action: anomaly detection, demand forecasting, menu intelligence and automated task tracking.
- 340+ Branches

- 340+ Branches

- 325+ Branches

- 200+ Branches

- 100+ Branches

- 95+ Branches

- 90+ Branches

- 85+ Branches
Tell us your needs yourself, get a quote quickly
Finding the right solution for your business is now very easy. Define your needs in just a few steps and receive your tailored quote right away.
- Identify your device needs
- Define your production points
- Choose your online platform integrations
- Get your tailored quote quickly
Needs Assessment Form
Define Your Device Needs
Touch Terminal
Touchscreen POS terminals for checkout points
Waiter Handheld
Mobile handheld terminals waiters use to take orders
robotPOS in Numbers
We lead the digital transformation of hospitality businesses
- Countries Served
- 10+
- Domestic Software
- %100
- Businesses Live
- 4000+
- Chain Brands
- 80+
- Team Members
- 100+
