The Center for Data Innovation recently spoke with Tim Stuart, CEO of AMP, a Colorado-based company using an AI to help waste management facilities recover recyclable materials more efficiently. Stuart explained how AMP’s system combines vision technology, sensors, and machine learning to automatically identify recyclable items and separate them from general trash at scale.
David Kertai: What does AMP offer?
Tim Stuart: AMP provides a fully integrated sorting system that combines conveyor lines, sensors, mechanical sorters, computer vision, and an AI model. These systems operate as standalone facilities or integrate into existing recycling infrastructure, giving waste operators a complete automated solution instead of a simple add-on. The system uses computer vision to identify each item on the conveyor in real time by shape, color, texture, and branding. Machine learning then interprets that visual data with sensor inputs to decide how to sort each object, continuously improving accuracy, handling unusual items, and optimizing equipment performance.
After identifying each item, the system directs materials into the correct stream or removes them from general waste. The system connects directly to every component, adjusting conveyor speeds, controlling material flow, and detecting jams as they occur. With the intelligence and mechanical sorters working as one coordinated system, facilities recover more recyclable material, reduce manual labor, and run their operations more efficiently.
Kertai: How do you train your AI-powered system to correctly sort materials?
Stuart: Each machine’s AI model trains on datasets that reflect real-world waste conditions, including large volumes of labeled images and sensor data that teach the system to recognize key features of recyclable materials, such as shape, texture, size, and branding. During operation, the computer vision system captures what the sensors and cameras detect, and the AI model logs how each item is classified. All of this data flows into AMP’s Cloud platform, which tracks accuracy and equipment performance under different conditions.
This creates a continuous feedback loop, where every sorting decision is captured, analyzed, and compared against expected outcomes. When the system detects errors or inefficiencies, the AI model adjusts recognition parameters, recalibrates sensors, and fine-tunes machine settings. Over time, the machines sort faster and more accurately adapting to changes in packaging or regulations while reducing the need for manual sorting.
Kertai: How do your customers use the collected data to improve their operations?
Stuart: Data from AMP’s machines gives customers real-time insight into how materials move through their facilities. Instead of only seeing the end result, operators can monitor the types of materials coming in and how the equipment is handling them. This visibility matters because day-to-day waste isn’t uniform, one day a facility might receive more cardboard, another day more plastics, and knowing the composition of incoming materials, how long it takes a system to process, or materials strain on equipment, enables managers to adjust equipment settings and staffing to match the workload required.
Customers can make immediate adjustments to reduce manual labor, improve uptime through predictive maintenance, and adapt to new packaging materials. Beyond daily operations, customers use detailed information about changing material composition and potential sorting bottlenecks to meet reporting requirements and plan long-term efficiency improvements.
Kertai: How do you ensure your system delivers consistent accuracy?
Stuart: We achieve consistent accuracy by designing the entire sorting line around a unified AI and sensor system, rather than retrofitting advanced tools onto older equipment. Every conveyor, camera, and sensor works in unison with the AI model, which provides real-time decision-making and precise control over each item’s movement.
We also continuously calibrate sensors and update algorithms to recognize new packaging and emerging patterns. These measures ensure our machines can accurately identify and sort over 90 percent of the material passing through them.
Kertai: What upcoming developments is AMP working on?
Stuart: We’re developing AI-driven machines that can sort both recyclable and organic materials, such as food waste, directly from mixed waste. By separating organics before they reach landfills, our systems increase industrial-scale composting, help communities meet sustainability goals, and reduce the volume of waste buried or burned. Over time, this AI-driven technology advances a more circular system where more materials are captured, reused, and kept out of landfills.
We’ve also signed a 20-year contract with the Southeastern Public Service Authority of Virginia to provide waste processing services and boost its recycling rate. Under this long-term partnership, our systems will be able to process 540,000 tons of waste annually, significantly strengthening the region’s waste-management capabilities.
