Californian shelf digitisation firm Simbe has improved its AI-powered computer vision (CV) technology for in-store intelligence.

These enhancements to Simbe Vision move beyond simply identifying problems to implementing AI-driven actions that optimise store operations with precision and efficiency.

“The most successful retailers are leveraging real-time, AI-powered insights to drive execution excellence,” said Brad Bogolea, CEO of Simbe. “What sets industry leaders apart is how they’re combining cutting-edge computer vision with Simbe’s Store Intelligence platform to identify the next best action for their teams. Together, we are enabling faster, smarter decision-making exactly when it matters most, creating a competitive advantage in today’s challenging retail environment.”

Simbe Vision enhances this technology with several capabilities, developed in partnership with retailers:

• Low-stock-on-shelf detection: Combines volumetric detections and depth sensing to identify products running low before they become out-of-stock, enabling proactive replenishment and preventing lost sales. Volumetric-based low-stock-on-shelf detection provides 1.4x more early signals about shelf condition compared with traditional methods.
• Shelf intelligence: Detects product spreads, plugs, misplaced items and missing price tags using AI technology, improving planogram accuracy and reducing inventory distortion.
• Inventory monitoring: Adapts to dynamic retail environments by combining mobile and fixed sensors. Computer vision seamlessly fuses data from autonomous robots and fixed cameras, eliminating planogram maintenance and capturing real-world shelf conditions.
• Product recognition and similarity detection: Employs AI-powered visual search for product identification across shelves, distinguishing even nearly identical products such as flavour or size variants to ensure correct placement and prevent mis-merchandising.
• Shelf-to-stock comparison: Cross-checks on-shelf inventory with backroom stock data to uncover hidden plugs that appear fully stocked at a glance, preventing phantom out-of-stocks and improving replenishment accuracy.
• Automated shelf tag verification: Uses barcode scanning, optical character recognition (OCR) and proprietary machine learning to detect mismatches between price tags and products, eliminating costly errors.

Simbe Vision delivers continuous, full-store visibility with 98.7% SKU-level identification accuracy and over 99.3% shelf condition recall; typical computer vision systems average 85 to 90% accuracy and 75 to 85% recall. Overall shelf condition precision exceeds 99%, ensuring retailers can trust the data and recommendations they receive.

Simbe has analysed over 60 billion retail shelf images to date, providing real-time data on inventory levels, pricing and product displays. The company expects to process twice as many images in 2025 as in 2024, helping retailers worldwide make quicker and more informed decisions.

“What sets Simbe apart is our ability to deliver full-store intelligence rather than limited audits,” said Jari Safi, vice president at Simbe. “Our deep learning algorithms provide SKU-level accuracy across our database of more than 18 million SKUs – the largest of its kind in retail – distinguishing even the smallest product differences to ensure precise inventory tracking and pricing integrity.”

Unlike most fixed-camera systems that rely on rigid planogram data, Simbe’s Tally Spot uses real-time insights from Tally’s daily store traversals. This removes the need for store teams to follow rigid workflows and delivers more adaptive, accurate shelf intelligence.

The computer vision system scans 5400 items per hour, providing real-time data that enables store teams to respond quickly to inventory issues. The system detects up to ten times more out-of-stock items compared with manual audits.

Simbe Vision doesn’t just surface problems, it recommends the next-best actions based on business impact, helping store teams prioritise what matters and shift from reactive issue handling to proactive execution that improves labour efficiency and drives real results.

Simbe Vision has been validated by global retailers including BJ’s Wholesale Club, Schnucks, Wakefern, Carrefour SA and SpartanNash, delivering measurable impact through sharper pricing accuracy and faster execution. The accuracy of the computer-vision technology empowers retailers to achieve compliance, fewer execution errors and transformative operational efficiency at scale.

Simbe’s platform includes the Tally autonomous, item-scanning robot, which identifies exact product location, stock level, and pricing and promotion information with computer vision. Simbe works with global brands in the USA, Europe and Asia. For more information, visit www.simberobotics.com.