AI Optimizes In-Store Product Placement
A platform from Auki is using phone-based store mapping and AI to optimize physical retail operations. By linking spatial data with sales figures, the system suggests optimal product placements to drive sales and generates tasks for staff, aiming to boost both labor productivity and inventory accuracy.
Hong Kong-based Auki Labs is the company behind the platform, founded in 2019 by CEO Nils Pihl. The firm is developing a decentralized spatial computing protocol known as the "posemesh," designed to create a shared, machine-readable understanding of physical spaces, effectively building a "web for the real world." The system bypasses the need for expensive infrastructure like beacons or high-end sensors. Instead, it relies on low-cost visual markers—stickers that cost only a few cents—and standard smartphone cameras to map a retail environment with centimeter-level precision. One 2,500-square-meter store was reportedly mapped for just $20 in about five hours. This technology underpins Auki's "Cactus" platform for retail. By creating a precise digital twin, the AI can optimize routes for order picking, potentially cutting the distance an associate walks by more than half. A case study with Stora Coop Visby noted that the system saves fifteen minutes per employee each day on task handovers alone. Auki's decentralized approach ensures that retailers retain sovereignty over their spatial and operational data, rather than feeding it to a centralized third-party platform. The protocol, an open and permissionless standard, allows various devices—from handheld scanners to autonomous robots—to share spatial reasoning data and collaborate within the environment. The company has reportedly raised between $15.6 million and $19 million and secured a multimillion-dollar contract with Sweden's largest grocery retailer. It also ran a pilot with Pepito's Supermarket in Bali to roll out customer-facing AR navigation, allowing shoppers to search for a product on their phone and follow augmented reality directions to its exact location. The core software development kit (SDK), called ConjureKit, provides the tools for building applications on the posemesh. It includes specific modules for advanced functionalities like real-time 3D hand tracking from a standard camera feed and sophisticated QR code pose estimation, enabling more interactive and aware AI-driven applications on the edge.
