The planet doesn't need more pledges. It needs proof.

Every year, another sustainability report lands in my inbox. Another net-zero pledge. Another commitment to circularity. And every year, the actual numbers barely move.

Recycling rates are flat. E-waste is growing. Brands put “recycled” on packaging and nobody can verify it. Consumers want to do the right thing and have no way to know what the right thing even is. Regulators write rules and cannot audit compliance. We have been having this conversation for twenty years, and the conversation is going nowhere.

The numbers are not subtle. The Global E-waste Monitor put global e-waste at 62 million tonnes in 2022, up 82 percent on 2010, with only 22.3 percent of it documented as formally collected and recycled. On current trends that share goes down, not up, while the total is projected to reach 82 million tonnes by 2030. Twenty years of circular-economy messaging, and the recycling rate is heading the wrong way.

This is not a willpower problem. It is a data problem.

Here is what I mean. When you throw something away, its history dies. Who owned it. What warranty covered it. Whether it was repaired. Whether it was actually recycled, or just shipped to a landfill with a paperwork story. None of that exists in a structured, verifiable form. So circularity stays a marketing claim — something brands put in reports, not something engineers can build on.

I am building Ourbits to fix this. Not another sustainability app. The infrastructure layer that makes the circular economy real. Here is how I think about the technical problem, because the problem is harder than people realise.

Every product needs a graph, not a barcode

The foundation is what we call a persistent product graph. Every product gets a graph-native identity that follows it from manufacture to end of life. Ownership changes, warranty claims, repairs, resale, recycling — every event gets recorded on the same object.

Why a graph and not a database? Because a product’s life is not linear. It branches. You buy a phone, sell it two years later, the buyer drops it, gets it repaired, sells it again, and eventually it reaches a recycling facility. That is not a row in a table. It is a chain of events across multiple owners, multiple products (the battery, the screen), and multiple lifecycle stages. A graph captures that. A barcode does not.

The graph is also what makes everything else possible. Without a persistent identity there is no object to attach history to, no history to verify, and no foundation for the layers above it.

Every event needs to be immutable

Once you have the identity, you need the history. We use event sourcing: every lifecycle event (a claim filed, a repair completed, a product resold, a recycling confirmed) gets recorded as an immutable event. Not a database row someone can edit. An event. Timestamped. Permanent.

This is what makes the system auditable. You can replay the entire history of any product and see exactly what happened, in what order, with what evidence. A regulator auditing a sustainability claim does not take the brand’s word for it. They replay the events.

Event sourcing also solves the trust problem. When a brand says “this product was recycled,” they are not making a claim. They are pointing to an event — recorded at the time of recycling, by the recycler, with cryptographic proof. That is not a pledge. That is evidence.

Agents, not dashboards

The system needs to do things, not just record things. So there is an agent layer. Three core agents, each with a specific job.

A ClaimAgent investigates warranty claims. It gathers evidence, builds the case, files it with the brand, and tracks the decision. When it hits something it cannot resolve — missing documentation, conflicting information, low confidence — it pauses and asks a human. Otherwise it runs end to end.

A CircularAgent manages the post-purchase lifecycle. It tracks buy-back eligibility, identifies products worth recovering, coordinates with brand buy-back programs, and routes returned items into resale, refurbishment, or recycling.

A ComplianceAgent knows what the EU’s Digital Product Passport requires and prepares it automatically. When regulation changes, the agent adapts. When a brand needs to prove compliance, the agent produces the evidence.

These are not chatbots. They are deterministic workflows with AI for the parts that need reasoning: extracting data from messy receipts, resolving ambiguity in claims, deciding when to escalate. Our stack is 80 percent deterministic infrastructure (event sourcing, state machines, rules engines) and 20 percent AI. The AI gets the demo. The infrastructure gets the trust.

Blockchain does one job

People hear “blockchain” and their eyes glaze over, or they assume crypto. We use blockchain exactly once, as a proof layer. Hashed references to lifecycle events, timestamps, verifiable recycling credits using zero-knowledge proofs.

The blockchain does not run the system. It makes the system’s records tamper-resistant and auditable. That is the only job it needs to do. No tokens, no crypto, no speculation. Just a notary for the physical world.

Why this matters

Right now, circularity is expensive and unverifiable. When a brand runs a buy-back program they cannot identify which products to target or verify their condition. When a recycler processes materials they cannot prove recovery. When a regulator audits a sustainability claim there is nothing to audit.

The circular economy does not fail because people do not care. It fails because the infrastructure does not exist.

When the data layer is real, everything changes. Brands run buy-back programs at scale because they can identify and verify products. Recyclers document recovery and trade credits. Regulators audit compliance in real time. And consumers — people who do not read sustainability reports and never will — participate automatically. They photograph a receipt, file a claim, sell a used product. The system captures their history whether they think about circularity or not.

The regulation is no longer hypothetical

Two of these forces are already binding in Europe.

The Digital Product Passport is the one with a clock on it. The EU’s registry for passports went live in July 2026, and from 18 February 2027 batteries in electric vehicles, e-bikes and e-mopeds, and industrial batteries above 2 kWh must carry a machine-readable passport tied to the physical unit. Iron and steel follow in 2026, then textiles, tyres and aluminium in 2027, furniture in 2028. Each product group gets its own delegated act defining the required data fields.

Extended producer responsibility pushes from the other direction, making brands financially responsible for what they put on the market and for what comes back.

For the product groups these rules cover, neither is a nice-to-have. They are conditions of placing goods on the European market, not reporting exercises. But both are phased, which is the window. The data they demand — who owned this, was it repaired, did it actually get recycled — is data almost nobody has been collecting. That is the gap, and it is the one being built into.