Amplio estimates that U.S. manufacturers hold or waste roughly $350 billion annually in unused equipment, spare parts, and other maintenance and operating inventory. Much of it is recycled, scrapped, or ultimately sent to landfill. The secondary market for industrial assets exists, but it is slow, opaque, and difficult to access at scale. For decades, that loss was simply accepted as a cost of doing business. Trey Closson spent a decade inside that world. He built Amplio with cofounder and CPO Taha Zinifi to fix it.

Early in Amplio’s life, one of its first customers put that problem in front of him directly. They were sitting on a significant amount of brand-new inventory, still in the box, and asked if Amplio could do something better than recycling it. Closson said yes, even though he did not yet know exactly how. Before that moment, Amplio had been building software that helped manufacturers monitor supply chain risk. Customers found it useful, but not useful enough to pay seriously for. That conversation changed everything.

The pivot that followed turned Amplio into a different kind of company. Today, the platform uses AI to appraise surplus industrial assets and determine the best path for each one, whether to redeploy, resell, recycle, or dispose of it compliantly. What traditionally took analyst teams several weeks can now be done in minutes, across inventory lists spanning tens of thousands of SKUs.

In this interview, Closson walks us through the AI behind the platform, the hard lesson his first product taught him about the difference between useful and essential, and why billions in industrial waste have gone unrecovered for so long.

The Problem

1. Most people do not think of surplus inventory as a crisis. Amplio has cited a $350 billion annual U.S. manufacturing waste figure tied to unused capital equipment and MRO. How did that problem stay hidden for so long, and why is it so hard for manufacturers to solve on their own?

Surplus inventory stays hidden because it accumulates gradually and never triggers an alarm. Manufacturers are laser-focused on uptime, so procurement operates on a “never run out” mandate. The result is systematic over-ordering baked into the culture: safety stock gets replenished even when demand has shifted, and no one is incentivized to flag the excess because the cost of a stoppage always feels greater than the cost of unused parts sitting on a shelf.

The deeper problem is a timing mismatch. Production lines evolve constantly — new equipment gets swapped in, processes get redesigned — but the MRO inventory ordered for the old configuration doesn’t disappear. It just becomes obsolete. Because these surpluses build up across dozens of cost centers and facilities, no single person owns the full picture. By the time leadership notices the carrying costs, the inventory is already years old and the original buyers have moved on. Solving it internally requires cross-functional coordination, specialized market knowledge, and a disposition process that most manufacturers simply aren’t set up to run.

2. Before companies like Amplio existed, what did a manufacturer typically do with a warehouse full of idle equipment, and what did that actually cost them in write downs, storage, labor, and missed recovery?

To this day, facility managers have three options, none of which feel good. The first is to do nothing, letting equipment sit in warehouses for years while inventory carrying costs quietly drain value. The second is to cut their losses and dispose of it — except “disposal” often meant paying a fee to haul it away, turning a write-down into an additional out-of-pocket cost. Last, opportunistic operators will bring in liquidators to bid on their inventory, but that tends to be only a partial solution as they’ll typically only take the best surplus and leave the facility with a still large surplus problem.

That’s not to mention the lack of options for corporate leaders, who often don’t even have visibility into the issue. To do anything about a surplus challenge, they first have to get all the individual facility managers to play ball — and then they face the same challenges the individuals do.

What Amplio Does

3. Give me the simplest version: a manufacturer sends Amplio a list of 50,000 SKUs. What happens next, and what does the manufacturer receive at the end?

The manufacturer sends over their surplus inventory list, and Amplio’s AI gets to work: cross-referencing those SKUs against market data, historical transaction prices, and demand signals to generate a valuation and optimal disposition channel for each item. Instead of a months-long manual appraisal process, the manufacturer gets back a clear, data-driven picture of what their surplus is actually worth on the secondary market in a fraction of the time.

From there, Amplio handles logistics and remarketing and connects that inventory to buyers who need it. The manufacturer ends up with recovered cash from assets that were materially hurting operations without having to build out the market expertise or buyer network to do it themselves.

4. Amplio says it helps customers decide what can be redeployed, what should be sold into secondary markets, and what should be recycled or disposed of compliantly. What drives those recommendations, and how does the AI weigh those options against each other?

The AI is essentially asking one question for every SKU: what’s the highest-value path for this asset? It starts with market data. If there’s real buyer demand and a strong recovery price, remarketing is the obvious answer. If the item has value internally across other facilities, redeployment comes first since that’s a dollar-for-dollar recovery with no transaction cost.

For everything else, we’re weighing residual market value against the cost and complexity of moving it. Items with thin resale markets or high logistics costs get routed toward auctions, recycling, or compliant disposal. The goal is to make sure nothing gets scrapped that could have been sold, and nothing gets sold for less than it costs to move.

How the AI Works

5. What is actually happening under the hood between raw spreadsheet ingest and a final disposition recommendation? Walk me through the AI workflow step by step.

Without sharing too much, Amplio’s AI breaks what would otherwise be an overwhelming problem into a series of discrete, manageable tasks handled by specialized agents working in concert.

An orchestration layer ties it all together, sequencing the agents, flagging low-confidence outputs, and applying guardrails to catch recommendations that don’t pass a sanity check. And when the stakes are high or the data is ambiguous, human experts can step in to review and refine before anything goes back to the customer.

6. Industrial inventory data is often messy, with incomplete descriptions, inconsistent part numbers, and condition data that varies by site. How does the AI handle that before it can appraise anything?

Clean data is the exception, not the rule. Manufacturers typically send over spreadsheets with inconsistent part numbers, truncated descriptions, missing manufacturer data, and formatting that varies site by site. Before any appraisal can happen, the AI has to make sense of what it’s actually looking at.

It’s rarely perfect, but the goal is to get every SKU to a confident-enough identification that a meaningful valuation can follow. The alternative (asking the manufacturer to clean the data themselves) would incur significant costs, exactly the opposite of what we deliver.

7. How much of the AI challenge is valuation, and how much is ontology, meaning understanding what an asset actually is across inconsistent manufacturer names, part numbers, and site level descriptions?

They’re inseparable, really. Ontology is the foundational problem, as you can’t price something you haven’t identified to some level.

The good news is that it doesn’t need to be perfect; it needs to be “good enough.” If the AI can confidently establish what category of asset something is and its general specs, that’s usually sufficient to generate a meaningful ROI estimate and, more importantly, make the right disposition call. Getting a valuation precise to the dollar matters less than correctly routing an asset toward remarketing versus recycling. The ontology work is what makes that possible.

8. Pricing surplus industrial equipment is hard because markets are thin and comparables are sparse. What distinguishes Amplio’s AI appraisal from a generic pricing model or a simple rules engine?

A generic pricing model might assign a flat recovery rate to a category — say, 20 cents on the dollar for all industrial valves — and call it a day. The problem is that within any given category, actual market value varies based on brand, spec, condition, age, and current buyer demand.

Amplio’s AI works from real comparables, applied at scale. Instead of broad category assumptions, it’s looking at what specific types of equipment have actually sold for in secondary markets and using that to inform each appraisal. The result is a much more granular and accurate picture of what an asset is worth. For a manufacturer with 10,000 SKUs, the difference between a generic recovery estimate and a comparable-driven one is thousands of SKUs directed to suboptimal disposition channels.

9. Liquidity seems just as important as price in your workflow. How does the AI estimate whether a given asset is actually likely to move in the secondary market?

Liquidity might actually matter more than price. A high appraisal on an asset that no one is buying is a trap. It just means the equipment sits longer and carries more cost.

Amplio’s AI incorporates real-time market analysis to assess that question for every asset, though the specifics of how we do it are proprietary. What we can say is that liquidity is a primary input into the disposition recommendation, not an afterthought. An asset with strong liquidity and a moderate price might be a better candidate for remarketing than one with a high theoretical value but a thin buyer pool. Getting that call right is what separates a genuinely useful appraisal from one that just looks good on paper.

10. At what point does the system defer to a human expert, and which categories still require the most human review?

Without getting into the specifics of how the system flags edge cases, the short answer is that every appraisal Amplio produces gets a human expert review before it goes back to the customer. That’s a deliberate choice, not a fallback. Enterprise manufacturers are making significant financial and operational decisions based on these recommendations, and human sign-off is what makes that trust possible.

Performance and Trust

11. How do you evaluate whether the AI got the appraisal right: recovery rate, speed, liquidity predictions, recommendation accuracy, or some other benchmark?

We evaluate against our track record of real transactions. When an asset goes through the full process and actually sells, that outcome becomes a data point we can use to assess how well the appraisal predicted reality: did the recovery rate land where we expected, did it sell in the timeframe the liquidity model anticipated, did we route it to the right disposition channel?

Over time, that feedback loop is what sharpens the model. Each completed transaction is essentially a grade on the prior appraisal, and that compounding data advantage is hard for a newer entrant to replicate.

12. Amplio has said current customers are seeing value recovery of up to 5x industry standards. What does the AI do that a traditional industrial liquidator with experienced appraisers cannot?

The 5x recovery figure is really a story about remarketing and our business model, not just appraisal. Amplio’s network of specialized affiliates means that each asset gets routed to the channel where it’s most likely to find the right buyer at the best price, rather than the most convenient one.

We also typically enter profit shares with our clients, which fully aligns our incentives. We make more when they make more. Other liquidators hunt buyouts, in which they bid low for great assets and then mark them up 10x to sell them.

13. How do you prevent bad recommendations in edge cases, such as equipment with sparse metadata, obsolete parts with no clear secondary market, or assets with compliance constraints?

No system gets every edge case right, and we’re upfront about that. Appraising secondary market value is genuinely hard, and the honest answer is that human review is the most important safeguard when the AI is working with limited signal.

But there’s also a structural protection built into how Amplio operates: our profit-share model means that if we undervalue an asset and it sells for significantly more, the client captures that upside. That alignment of incentives is its own kind of guardrail: clients can trust that the recommendations are made in good faith even when the data is imperfect.

Learning and Improvement

14. What feedback loop from completed sales or redeployments most improves the model over time: realized price, time to sale, buyer behavior, or project level execution data?

All of it matters: realized price, time to sale, buyer behavior, and how projects actually execute end to end. Each completed transaction is essentially a report card on the prior appraisal, and every data point contributes something different: price tells you if the valuation was accurate, time to sale validates the liquidity prediction, buyer behavior reveals which channels and networks are performing, and project-level data captures the full picture.

The specifics of how we weight and integrate those signals are proprietary, but the compounding effect is real.

What Comes Next

15. The Series A is partly earmarked for agentic system development. What is one capability you can talk about today that will make Amplio’s AI materially more autonomous than it is now?

The biggest unlock is orchestration across the entire value chain, not just appraisal. We offer the appraisal for free because the real value is in what happens after. What we’re building toward is an agentic system that coordinates the full liquidation process end to end: identifying assets, routing them to the right channels, engaging the right buyers, and managing execution — with minimal human intervention required to move between steps.

Right now there are handoffs. The vision is a system where those handoffs increasingly happen automatically, with agents managing the sequencing and humans stepping in only where judgment is genuinely required. That’s what makes Amplio something fundamentally different from a software tool that appraises assets — it becomes the operating infrastructure for the entire surplus lifecycle.

Origianl Creator: Genaro Palma
Original Link: https://justainews.com/industries/industrials-and-manufacturing/amplios-ceo-on-closing-the-loop-in-industrial-supply-chains-with-ai/
Originally Posted: Mon, 23 Mar 2026 11:26:54 +0000