In a major advancement for the global recycling industry, TOMRA Recycling has officially unveiled a breakthrough AI-native platform developed by PolyPerception, alongside the introduction of three new deep learning applications for its award-winning GAINnext™ technology. Unveiled at the IFAT 2026 trade fair in Munich and the PRSE exhibition in Amsterdam, these innovations underscore the company’s relentless focus on advancing AI-driven sorting solutions. Coinciding with these technological launches, TOMRA announced a significant strategic move by acquiring a 51% majority stake in PolyPerception, effectively closing the loop between real-time data analysis and physical sorting operations.
From Reporting to Interpreting: The New AI Agent for Recycling
The newly introduced AI-agent platform by PolyPerception represents a massive evolutionary step for its existing Waste Analyzer—an AI-powered waste analytics solution designed to improve sorting performance through comprehensive, end-to-end material tracking. The most significant breakthrough of this updated system is its natural language interface.
Facility operators now have the unprecedented ability to “chat” directly with their plant data using plain language. For example, a user can simply ask, “How did changing the settings on the recovery line affect our purity?” Operating with artificial intelligence at its core, the platform inherently understands the operational context and immediately generates natural language answers supported by detailed data breakdowns. This effectively removes the traditional technical barriers between complex data spreadsheets and rapid operational decision-making.
Furthermore, while traditional AI tools in the waste management sector are typically limited to “reading” and reporting historical data, this new platform boasts “writing” capabilities. Acting as an active agent within the facility rather than a passive observer, it utilizes its deep domain knowledge of the recycling process to autonomously create custom quality reports and set critical operational alerts within seconds.
“With the introduction of our new agent-based platform, recycling plants now gain a new cognitive layer,” stated Nicolas Braem, CEO and Co-Founder of PolyPerception. “Data is no longer just reported – it is interpreted, explained and transformed into relevant insights in a few seconds. Operators can interact naturally with their plant, ask questions, explore material behavior and receive clear, actionable answers in real time.”
Open Data Integration and Advanced Search Features
Providing full operational transparency, the groundbreaking technology allows recyclers to integrate the platform’s data directly into their existing facility management systems. Consequently, plant managers can seamlessly query waste statistics or specific material purity levels through their own centralized dashboards, eliminating the need to log into a separate, isolated system.
To help facilities rapidly respond to constantly changing material streams, the platform introduces two powerful new search methods:
-
Similarity Search: Operators can simply right-click on a problematic object—such as an electronic vape—to instantly identify every other visually similar item passing through the stream. This feature is absolutely critical for rapidly spotting severe fire hazards, like hidden batteries, without the time-consuming process of training an entirely new AI model.
-
Text and Brand Search: Users are empowered to search for specific consumer brands or object types, such as “filled refuse bags” or “diapers,” allowing them to monitor exactly what is passing through the sorting facility in real time.
Expanding the GAINnext™ Ecosystem: Three New Applications
To perfectly complement this digital progress, TOMRA has launched three new deep learning applications specifically designed for its GAINnext™ ecosystem. These solutions are engineered to eliminate long-standing industry bottlenecks where traditional sensor-based sorting technology has historically reached its absolute limits.
-
Food-Grade PET Trays: Addressing the surging market demand for food-grade PET, this application recognizes that tray material is becoming a critical new feedstock alongside standard bottles. By aggressively training the GAINnext™ system on thousands of images, the AI can now accurately distinguish between supermarket/takeaway trays and medical/consumer packaging based strictly on shape and intended use. Achieving purity levels of over 95%, this breakthrough proves that PET tray sorting is now a highly viable commercial business case rather than a technical impossibility.
-
High-Precision Copper Recovery: In the heavy metals sector, TOMRA is introducing an application specifically for “copper meatballs,” directly supporting a steel market transitioning toward decarbonization. The advanced GAINnext™ technology automatically identifies complex copper-steel composites—such as motor armatures—even within heavily oxidized or exceptionally dirty material streams. This outstanding selectivity allows metal recyclers to successfully upgrade standard rebar-grade scrap into premium, high-value furnace feedstock.
-
UBC Aluminum Recovery: The third addition is a massive, high-throughput solution tailored for used beverage can (UBC) aluminum recovery from mixed packaging streams. Having been successfully launched in North America, the system is now adapted for the European market. The GAINnext™ UBC application delivers up to 33 times more throughput than manual human sorting while consistently achieving 98% purity or higher. By instantly detecting and automatically ejecting non-UBC materials, the technology provides a vastly more efficient pathway for closed-loop aluminum can-to-can recycling.
A Technology Turning Point
Lars Enge, EVP and Head of TOMRA Recycling, emphasized that while artificial intelligence has always been embedded in the company’s DNA, the industry is entering an entirely new phase. “With our acquisition of a majority stake in PolyPerception, we are moving beyond AI as a sorting tool to AI as a central intelligence for the recycling plant,” Enge explained.
He concluded that these combined launches signal a true technological turning point for the waste sector. Deep learning is no longer just tackling complex sorting challenges; it is actively linking deep insights directly to physical action across the entire plant. By moving beyond high-speed detection toward a new era of intelligent, connected sorting, TOMRA continues to translate the world’s most advanced AI into measurable, real-world value for its recycling customers.
Reference and Official Source:
-
Official TOMRA Press Release (May 07, 2026): TOMRA launches next-generation AI platform and expands GAINnext ecosystem
