Banner promoting IARC 2026 (International Automotive Recycling Congress), 25–27 March 2026 in Hamburg, Germany, with the tagline “One ticket. Full access to congress, exhibition & plant tours” and a “Register now” button.

Why Agentic AI Will Expose the Yards That Aren’t Ready

The next wave of AI doesn’t answer questions; it takes action. And it cannot act inside an operation that isn’t in order.

Lisa Samuel says agentic AI will expose rather than fix weak auto recycling operations. Because this next wave of AI acts across real processes, yards with poor data, fragmented tools, weak security and undocumented workflows will struggle, while disciplined businesses with clear systems will be better placed to benefit.

Lisa Samuel explains why agentic AI will expose weak auto recycling operations. — Lisa Samuel

Lisa Samuel, Founder of PayBuddy and Sabhi, explains why the next wave of agentic AI will not simply reward the most technologically ambitious auto recyclers. Instead, it will expose yards with weak processes, poor data, fragmented systems, and unresolved operational fundamentals, making readiness less about AI itself and more about discipline, structure, and execution.

You can drop the most advanced engine in the world into a car, but if the frame is rusted and the brakes don’t work, you haven’t built a race car. You’ve built a faster accident.

That is more or less what is happening as artificial intelligence arrives in auto recycling.

Walk through a dozen operations, and you’ll find the same things over and over: a website running without basic security, payments taken in ways that invite a chargeback, no real defense against fraud, and daily processes that hold together only because someone remembers how they’re supposed to go. These aren’t exotic problems. They’re the basics, and across much of the industry, the basics still aren’t handled.

AI isn’t going to solve these issues. It’s going to expose them.

Yet the loudest conversation in the business right now is about none of that. It’s about AI. The pitch is everywhere: add intelligence to the yard and watch it transform, and it is reaching operators who haven’t yet secured their own website or closed the gaps quietly costing them money every month. Before a recycler spends a dollar on AI, the harder question is whether the fundamentals are in order at all. Because the version of AI actually coming carries a requirement the sales pitch never mentions, and an operation that hasn’t handled the basics is the one most likely to get nothing from it.

The shift from answering to acting

Most of what gets called AI today is the kind that answers: you ask, it responds. It drafts, summarizes, replies. Useful, and fundamentally passive, it produces words and waits for a person to act on them. That is the chatbot nearly every vendor is selling, and it is a feature, not a transformation.

The shift the technology world is actually focused on is different, and it has a name: agentic AI. Where a chatbot answers a question, an agentic system pursues a goal, gathering information, making decisions, and taking a sequence of actions across multiple steps with limited human intervention. Traditional AI answers questions. Agentic AI executes processes. The difference is between a tool that tells you what to do and a system that goes and does it. Ask a chatbot to help plan a trip, and it hands you a list to go book yourself; give an agentic system the same goal, and it works toward the outcome, handling the steps in order rather than handing the work back to you. That is the leap the whole field is making, and it is why the people building this technology are far less impressed by chatbots than the market is. They know the chatbot is the shallow end.

Why acting requires a foundation that answering does not

Here is the part that decides everything, and it is easy to miss. A chatbot can sit on top of a messy operation and still work, because all it has to do is talk. An agent has to act, and to act reliably, it must follow the steps of a real process and trust what it finds at each step. That is a far higher bar. A business that runs on scattered tools and undocumented habits, where the same task is done five different ways depending on who is working that day, simply cannot clear it. The agent has nothing dependable to act upon. It fails not because the intelligence is weak but because the ground beneath it is.

The wider market is proving this the expensive way right now. Across industries, organizations are learning that you cannot simply bolt AI onto a disconnected operation and expect it to work, because a system built to take action is useless when the information it needs is not where it can reliably find it. Early enterprise data already suggests many agentic AI initiatives will fail, not because the models fail, but because organizations tried to run them on top of operations that could not support them.

Building the agent turns out to be the easy part. The operation underneath is where these projects die.

The operation underneath

So look honestly at the operation underneath a typical yard. Ask an operator how many separate tools their team relies on in a day, and most undercount, because they have stopped noticing. The tools accumulated one at a time over years, each added for a good reason, none of them chosen as part of a plan. Nobody set out to run a business this way. It is simply what happens to any operation that grows over time and solves each new problem as it appears.

The cost of that shows up not as a line item but as friction, the ordinary, daily kind that everyone has stopped noticing because it feels like just how the work goes. Information that exists but takes time to track down. Work that stalls because it depends on a single person who knows how to do a particular task. The same details handled inconsistently depending on who is doing them. None of it looks like a crisis on any given day. It looks like normal busy. And it is precisely this kind of friction, invisible, constant, and scaling with the business—that an agent cannot work around, because an agent has none of the memory and judgment your people use to paper over the gaps.

You cannot automate a process that only really exists in someone’s head.

Buying features instead of outcomes

There is a financial trap inside all of this worth naming plainly. In periods of fast technological change, the most common mistake a business makes is buying for today’s feature set instead of tomorrow’s needs. Much of what is being marketed as AI right now – chat, drafting, summarizing, basic automation – is commoditizing so quickly it will be ordinary, expected functionality within twelve to twenty-four months, not a durable advantage. The intelligence itself is becoming background utility, like electricity.

Paying a premium for it is paying for something that will soon be standard.

Technology markets have repeatedly followed this pattern. What is scarce today becomes expected tomorrow. Competitive advantage rarely lasts long from owning the technology itself; it comes from building an operation that can use it better than anyone else.

So the question a recycler should ask of any AI product is not “can it draft an email, answer a call, or answer a question.” Those capabilities are becoming free. The question is whether the investment improves the operation underneath. Those outcomes hold their value no matter which model wins next year. If a process is already slow and scattered, adding an AI chatbot does not fix it. It just gives the same process a new interface.

Where this leaves us

Agentic AI is real, and it is coming to this business. But it will not fix slow, scattered, inconsistent processes; it will only run on top of an operation solid enough to support it, or fail trying, as so many initiatives in other industries already do. The recyclers who come out ahead over the next decade will not be the ones who bought the most intelligence. They will be the ones who put their operations in order first, so that when intelligence became infrastructure, they had something for it to stand on. The advantage belongs to the businesses that can move information into decisions and decisions into action, which has surprisingly little to do with AI, and almost everything to do with the work done before it arrives.

About the Author

Lisa Samuel is the founder of PayBuddy and Sabhi. She has spent more than two decades working at the intersection of payments, commerce, risk, and technology, helping businesses navigate operational and transactional complexity.