AI and the circular economy: how to choose the first use case

A practical way to move from an interesting technology to a decision worth improving.

A company has a growing stock of returned equipment. Some units can be repaired, some can provide reusable parts, and others are suitable only for material recovery. The team has limited time to inspect each unit, and the information it needs is scattered across service records, photographs and spreadsheets.

Should it introduce AI? Perhaps. But that is not the first question. The first question is which decision is losing value today, and why?

AI can help interpret images, documents and operational data. It cannot, on its own, create a repair process, a market for recovered parts or reliable source data. The strongest starting point for an AI and circular economy project is therefore a specific operational decision, not a list of possible tools.

1. Find where value is lost

Start by following a product or material through its current journey. Where is useful value lost because a team lacks information, time or a viable route to keep it in use?

The answer may be at the design stage, in production, during maintenance or when a product returns. It may involve materials that are mixed, parts whose condition is unknown, or products routed to recycling before reuse has been considered. Circularity is broader than processing waste: it also means keeping products and materials in use at high value, as the Ellen MacArthur Foundation’s circular economy principles make clear.

Describe the loss in operational terms. Which product or material stream is involved? How much of it is affected? Who makes the decision now? What happens when the decision is delayed or wrong? If the team cannot answer these questions, it is too early to choose a technology.

2. Name the decision to improve

Turn the problem into a question that someone must answer in their daily work. For the returned equipment in our illustrative example, the question might be: which units should be repaired, dismantled for parts or sent to material recovery?

Better information could help the team make that choice faster and more consistently. AI might extract details from repair histories or assist with image-based inspection. But first compare it with simpler alternatives: clearer inspection criteria, better data capture, or a basic rule may already deliver most of the benefit.

This is a useful discipline. If the proposed output will not change a decision or an action, it is not yet a circular business case.

The route itself may need examination before introducing AI. Our article on repair, reuse and recycling decisions looks at that choice in more detail.

3. Check the evidence you can actually use

Before commissioning a pilot, inspect the data behind the decision. Are product identifiers consistent? Are condition and repair outcomes recorded? Can records be linked to the items they describe? Are the people running the process allowed to access and use them?

Perfect data is not required. A bounded pilot can also reveal what needs to be measured. But missing or fragmented information must be treated as a project constraint, not hidden behind a model. The European Environment Agency notes that some aspects of circularity remain poorly monitored because the relevant data flows are fragmented or absent.

Decide who will validate the source data, review uncertain outputs and maintain the process after the test. Without that ownership, even an accurate prototype is unlikely to become useful in operations.

4. Design a pilot that can produce a decision

Limit the first test to one product family, material stream or facility. Record the current process before changing it: time per assessment, decision consistency, share of products routed to each pathway, recovery value and any relevant environmental measure. Select only the metrics that fit the actual problem.

Then test whether the new approach improves the decision compared with that baseline. Include the cost of collecting data, checking AI outputs and changing the workflow. A faster classification is not a success if it sends more reusable products to lower-value recycling, or if the process is too expensive to operate.

Agree in advance what would justify scaling, revising or stopping the pilot. A useful test is one that supports a decision, including a decision not to deploy AI.

5. Put the people and the route to market around the tool

Circular outcomes depend on more than identification. Operations must be able to act on the recommendation; quality teams may need to approve reused parts; procurement or commercial teams may need a buyer or an internal application for recovered materials.

Bring these people into the project before the pilot starts. Otherwise, the tool may identify an opportunity that the organisation cannot use.

The first deliverable should be a short project brief: the value loss, the decision to improve, available data, process owner, baseline, pilot scope and criteria for success. Only then should the team decide whether AI is the right means to deliver it.

From the first question to a practical project

Managers who want to learn how to frame and evaluate these opportunities can explore the AI & Circular Economy Lab for Managers, developed with Federmanager Academy. Organisations with a specific product, material flow or business challenge can work with Tondo Lab to assess the opportunity and design a measurable pilot.

The aim is not to find somewhere to use AI. It is to find a circular decision worth improving, then test the most effective way to improve it.

Francesco Castellano

Francesco Castellano is a seasoned business leader and strategist with over 20 years of experience spanning research, finance, consulting, and entrepreneurship. He has held impactful roles, including serving as a consultant at Bain & Company, launching Uber operations in Turin, and working as Managing Director of a Swiss start-up. In recent years, Francesco Castellano founded Tondo, a hub of... Read more

Francesco Castellano is a seasoned business leader and strategist with over 20 years of experience spanning research, finance, consulting, and entrepreneurship. He has held impactful roles, including serving as a consultant at Bain & Company, launching Uber operations in Turin, and working as Managing Director of a Swiss start-up.

In recent years, Francesco Castellano founded Tondo, a hub of organizations dedicated to promoting Circular Economy approaches and supporting companies in transitioning to sustainable and circular practices. He is also the ideator and coordinator of the Re-think Circular Economy Forum, a high-profile event held across Italy to showcase innovative Circular Economy solutions.

Francesco Castellano collaborates with European institutions, serving as an expert for the European Commission’s Circular Cities and Regions Initiative and mentoring startups in the European Institute of Innovation and Technology’s (EIT) New European Bauhaus Booster Program. Through these roles, he actively supports the development and scaling of circular economy ventures across Europe.

He is also a sought-after speaker and lecturer, sharing his expertise on Circular Economy, Innovation, and Entrepreneurship at universities and international events. Francesco holds executive education certificates from prestigious institutions such as MIT, Harvard, and the University of Virginia, further solidifying his credentials in strategy, sustainability, and innovation.

Fluent in Italian, English, and Spanish, Francesco Castellano combines his diverse skill set with a passion for Circular Economy, Cleantech Innovations, and Entrepreneurship. His strong background in Corporate Strategy, Sustainability, Innovation Development, and Finance enables him to drive impactful change in every initiative he undertakes.

Condividi