Around 2 a.m., the compressor in a restaurant’s walk-in cooler starts drawing more power than it should. Its own agent notices the drift and checks the warranty. The owner’s agent already knows there’s a hundred-person private event on Saturday and that losing the cooler before then isn’t an option. Between them, they send a parts request to the distributor with a Friday deadline and the reason attached.
For once, the distributor knows why the part is needed and what happens if it’s late. The manufacturer, a step further back, learns that this compressor model is drifting in humid kitchens, months before a single warranty claim would have told it anything.
My recent conversations with companies that are reaching beyond the front end AI agent usually take the shape of the restaurant scene above. Nobody at the restaurant filed a ticket, and nobody at the distributor had to call and ask how urgent it was. The kind of information that usually gets lost, or gets relayed badly by three different people, moved with the request on its own. This isn’t an exclusive AI narrative, per se, but the right agentic AI deployments will reflect this scenario and begin to realign the entire value chain.
In The Edge of the Enterprise I argued that your own personal agent gives Customer Experience systems the earliest read on demand in the company. Once customers and products carry agents of their own, that kind of context will start to move through the rest of the supply chain too, and I think the consequences reach further than most CX conversations currently allow for.
Historically, each tier of the supply chain has seen only the order from the tier in front of it. The distributor sees a purchase order, the manufacturer sees the distributor’s replenishment, and the component supplier sees a forecast built on top of both. Every handoff loses some of the reasoning. Industries spent decades carrying safety stock, prepositioning parts, and deploying field service technicians against a plan with little context. I spent a long stretch of my career working with customers deploying service software, and I rarely saw a service record travel beyond the company that created it. The ticket lived in one system and the purchase order in another, and the supplier never saw either one.
Agents on both ends change what can make the trip. The customer’s agent brings intent and constraints, and the product’s agent reports its own condition and usage, which is far more reliable than anything pieced together from a service ticket. We’ve had connected products for more than a decade, and most of their telemetry ended up on dashboards nobody looked at after the first month. An agent, on the receiving end, can act on the reading. It can decide the drift matters, check the warranty, and ask for a part with the evidence attached. When that context moves with the request, the upstream parties get the reasoning they’ve always had to guess at.
Manufacturers have spent enormous sums trying to learn how their products behave in the field, and distributors have built whole planning practices around estimating urgency. In an AI agentic future, a good share of that information will show up with the request.
This changes where customer experience actually gets shaped. Whether the restaurant’s event goes well depends on a distributor and a manufacturer the owner has never spoken to, and on whether the context reached them intact. CX leaders who have focused on their own front line will increasingly find their customers’ experience being shaped two or three tiers away, by companies they have no direct relationship with.
None of this happens by default. The owner’s agent decides what gets shared and under what terms, and it has no obligation to tell a distributor anything. This is where protocols like MyTerms help establish enforceable contracts to share such critical data. In theory, going back to the EDI days, sharing critical information has been at the heart of modern supply chain management. Missing from that conversation has been the untapped CX data. It has been missing because this data was just outside of what could be captured, permission, and transmitted upstream. Agentic AI will significantly increase the impact that customers and products have on the entire supply chain. CX, in this world, is not an end point, rather a critical element to driving greater value across the entire workflow.
So trust starts to behave like a supply chain asset, one that determines how much any given company gets to know about the demand it serves. Being the largest supplier in the chain won’t guarantee much of it. The agents will share with whoever has earned it, and I suspect most supply chain leaders haven’t yet thought about what earning it looks like. I’d like to think CX leaders will be the ones who show them.