13 June 2025 | Expert Insight
Jari Safi, Vice President of Artificial Intelligence at Simbe, in an interview with Robotics Business News, shares how the company’s latest innovations are transforming retail through advanced robotics and AI. The recent upgrades to the Simbe Vision platform include low-stock detection, automated shelf tag verification, and real-time shelf-to-stock comparisons, enabling faster and more accurate in-store decision-making. By combining autonomous mobile robots with fixed camera systems, Simbe provides continuous and precise shelf monitoring across the entire store. Safi highlights measurable improvements seen by retailers such as BJ’s Wholesale and CarrefourSA, including fewer stockouts, quicker fulfillment, and stronger profit margins. Looking ahead, he outlines Simbe’s plans to expand its platform to address broader retail challenges such as monitoring product freshness, reducing food waste, and creating personalized shopping experiences using real-time shelf data.
What inspired the latest upgrades to the Simbe Vision™ platform, and how do they align with evolving retail challenges?
The latest upgrades to Simbe Vision were driven by the growing need for faster, more precise decision-making on the sales floor. Retailers increasingly want more than shelf visibility—they need actionable, strategic insights that help teams focus their time and attention where it matters most. Our latest iteration of Simbe Vision delivers exactly that.
One of the key new features is low-stock detection, which spots when products are running low on the shelf—preventing stock outs 1.4x earlier than traditional methods. Through our best-in-class computer vision, Tally is able to give store teams a heads-up so they can restock items in time and avoid missed sales.
Additional upgrades include automated shelf tag verification to reduce pricing errors, shelf-to-stock comparison to uncover hidden out-of-stocks, and enhanced recognition and similarity detection of near-identical product variants. These enhancements reflect how our industry-leading computer vision continues to evolve to meet retailers’ most pressing in-store challenges—transforming shelf data into next-best actions that improve execution and reduce excess manual work.
How does combining data from autonomous robots and fixed cameras improve the precision and scale of shelf monitoring?
Simbe is the only retail technology company to combine computer vision, robotics, RFID and fixed sensors in one platform—a combination that is crucial for delivering inventory accuracy and precise shelf execution at scale. Tally plays a critical role in capturing comprehensive store-wide data––analyzing tens of thousands of products with accuracy and efficiency. Fixed sensors, like Simbe’s Tally Spot, complement this coverage by providing real-time visibility in targeted, high-priority areas of the store. This includes high-traffic zones like fresh, rotisserie, and prepared foods, as well as high-risk sections like alcohol, cold medicine, checkout lanes. By combining mobile and fixed data sources, retailers gain a more granular, always-on view of shelf conditions––enhancing their ability to optimize stock levels, reduce shrink, and respond faster to operational needs.
Simbe claims 98.7% SKU-level accuracy—what innovations enabled this level of performance, and how is it validated across different store environments?
Simbe is widely recognized as a leader in retail shelf intelligence—and for good reason. Powered by cutting-edge computer vision technology that leverages a vast dataset alongside deep learning models, Simbe Vision is capable of detecting even the most subtle differences between SKUs, enabling near-perfect accuracy in inventory and pricing analytics. The system also reaches 150% further than competitive solutions, driving a level of precision in analyzing even the tallest store shelves. This ultimately empowers retailers to reduce execution errors, improve compliance, and deliver meaningful operational efficiencies at scale.
Validated in real-world environments across a range of retail formats––from grocery to wholesale, farm supply, and home improvement––Simbe’s technology is proven to consistently unlock market-leading performance, equipping retailers with the insights they need to improve operations and make smarter, data-driven decisions.
With over 60 billion shelf images processed, how is Simbe scaling its AI infrastructure to keep up with growing customer demand?
At Simbe we’re pioneering the evolution of AI from purely digital applications to embodied intelligence––developing purpose-built robotics for the physical world. Our technology is designed to seamlessly operate in any retail environment, setting a new standard for inventory management and human-robot collaboration.
To support the rising demand for retail intelligence, we’re also continuously optimizing our AI infrastructure for both performance and scalability. This includes reducing compute requirements through more efficient model architectures and by consolidating multiple tasks into single models. We’re also shifting more processing from the cloud to the edge, allowing each Tally robot to directly contribute to system scalability while maintaining fast, high-accuracy performance in-store.
Simbe has led the way in introducing robotics into dynamic, consumer-facing environments like retail stores. As adoption increases, we’re scaling to handle even greater volumes of data—aiming to double what we processed last year—while continuing to deliver real-time, actionable insights with unmatched precision.
What measurable outcomes have partners like BJ’s Wholesale and CarrefourSA experienced from deploying Simbe Vision™?
Simbe powers in-store intelligence for the largest number of retailers in the world. We have an established track record of delivering measurable results for leading retailers like BJ’s Wholesale, as well as Albertsons, Wakefern, Schnucks, SpartanNash, and more. On average, our retailer partners who have deployed Tally at scale report 60% fewer out-of-stock items, 90% fewer pricing and promotion errors, and 50% faster online order fulfillment. These gains directly support retailers’ bottom line, driving a more than 2% annual sales lift, a reported 4x ROI in just 3 months, and $15,000 in incremental margin per store, per month. These results are a direct outcome of the precision and scalability of our technology––and with continued advancements to Simbe Vision, we anticipate this impact to grow even further.
How does the system handle constant changes in store layouts or product assortments without relying on rigid planograms?
Tally is designed to effortlessly adapt and glide through any store environment, even as layouts or product placements change. Unlike competing technologies that rely on time-consuming and often outdated planograms, Tally maps stores’ current state in near real time––creating a “realogram” that provides accurate visibility into product availability, location, and pricing. Equipped with advanced path planning and obstacle avoidance, Tally operates smoothly even when relocated. Its slim shape and considerate maneuvers allows Tally to turn in place, navigate around obstacles, give customers space in crowded aisles, and move through tight spots—making it the most agile and customer-friendly robots available today.
In what ways does the platform support labor efficiency or help store teams reprioritize daily operations?
Historically, store teams have been responsible for manually auditing tens of thousands of products—a time-consuming, repetitive task that’s consistently ranked by associates as their least favorite. Tally transforms this process by autonomously scanning store aisles multiple times a day, compressing what once took days of manual effort into just a few hours. Tally then analyzes the data it collects in real-time to equip store teams with actionable, prioritized inventory tasks—enabling associates to easily spot and address inventory issues before they arise.
Store teams have overwhelmingly embraced Tally: 90% of store managers say they wouldn’t want to do their jobs without it. Not only does Tally eliminate one of the worst responsibilities, it also saves store teams up to 50 hours per week—freeing up valuable time for higher-impact work like customer service and hands-on store management.
Tally also improves accuracy in critical areas like pricing, promotions, and ecommerce fulfillment. In fact, stores using Tally for pick path optimization have cut order fulfillment times by as much as 50%. In a tight labor market, Simbe’s automation tools help store teams work smarter—not harder—improving outcomes for both employees and customers.
What’s next for Simbe in terms of AI and robotics—are there plans to expand beyond shelf intelligence into other areas of retail operations?
We’re already exploring how our computer vision platform can be applied beyond shelf intelligence to monitor things like product freshness and stock rotation, helping retailers minimize food waste and better protect margins. But that’s just the beginning.
Over time, the data foundation Tally is building will inform everything from planogram design and dynamic pricing to automated replenishment and last-mile fulfillment––turning the store into a living ecosystem that continuously adapts to each community, and, ultimately, to each shopper. Imagine recipe suggestions that reflect what’s actually in stock or targeted offers tied to emerging local demand. In short, we’re focused on expanding Simbe’s robotics and AI capabilities to unlock a more seamless, personalized, and efficient retail experience––from the shelf to the home.
