We Already Have ERP, MES, PLM

American industry has spent decades and hundreds of billions of dollars digitizing itself, so the objection is fair on its face. If coordination were really the problem, wouldn't the enterprise software already own it? A manufacturer runs an ERP for purchasing and accounting, an MES on the shop floor, and a PLM holding every revision of every model. Ask a skeptical CIO what a coordination layer adds to that and you should expect a short answer.

Then walk down the hall to whoever handles quotes. The RFQ arrived as a PDF attached to an email. The quote is going back in a spreadsheet. The new supplier is being qualified from scratch, over the phone, essentially the way it was done thirty years ago. Hundreds of billions of dollars of software, and the seam between two companies is still a person retyping numbers. Claim: ERP, MES, and PLM optimize execution inside the enterprise and cannot produce external coordination — shared supplier discovery, portable trust, structured demand intake across parties that have never met — and the proof is that procurement still runs on email, PDFs, and spreadsheets despite the enterprise stack; the tell of the gap is the share of inbound work that arrives unstructured and the time it still takes to qualify an unknown shop.

The category error is treating internal orchestration and external coordination as the same problem. One runs a company. The other runs the space between companies, which is exactly where the country's coordination failure lives.

The Objection at Its Strongest

The objection deserves its strongest form, because these systems are genuinely powerful and genuinely everywhere. In Gartner's own definitions, an ERP delivers "an integrated suite of business applications" across "finance, HR, distribution, manufacturing, service and the supply chain"; an MES manages "the execution of real-time, physical processes involved in transforming raw materials into intermediate and/or finished goods"; and a PLM manages products "through the stages of their life cycles, from concept through retirement." These are not toys, and their vendors are not small. PTC, which makes the Windchill PLM, reported roughly $2.3 billion in fiscal 2024 revenue, and it sits alongside SAP, Oracle, Siemens, and Dassault in a set of categories worth tens of billions a year. If ubiquity and budget were enough, we would not be having this conversation.

The Category Error

Read those definitions again and the boundary is sitting right there in every one of them. Finance, HR, the shop floor, the product lifecycle — all of it inside the four walls.

A PLM holds design intent beautifully and cannot emit a quote-ready package to a shop it has never heard of. An MES synchronizes execution on machines the company owns and does nothing whatsoever to qualify a supplier the company does not. An ERP manages material masters, work orders, and payments after a relationship and a record already exist, which is the tell: by the time the ERP has something to do, the external coordination already happened somewhere else, and that somewhere else was almost certainly an inbox. These systems assume the counterparty. Finding and trusting the counterparty is the actual problem.

The evidence that the seam is still manual is not anecdotal. Take the invoice — the most standardized, highest-volume, most regulated document in all of B2B. By Billentis's accounting it is still only about twenty-nine percent electronic worldwide, which means roughly seventy percent still moves as paper or an unstructured PDF. If the invoice is mostly manual with regulatory mandates pushing it, the RFQ and the quote, which have no such mandate, are worse. Deloitte's 2025 survey of chief procurement officers named the single largest barrier to delivering value: "siloed ways of working," at fifty-seven percent. No internal system dissolves a silo that spans two companies. It was never built to reach that far.

The Governed Design

The coordination layer sits deliberately outside the ERP, as Paper 23 argued, and interoperable with it rather than a replacement for it, as Paper 24 required. It carries the three things the internal systems structurally cannot: a structured Request that any buyer can send and any supplier can receive, a portable identity and performance record that survives crossing a firm boundary, and an intake that turns a messy email into a routable object before a work order exists anywhere.

Then it hands off and gets out of the way. Awarded work feeds into the winner's ERP and MES through documented APIs, and the shop still runs the job on its own systems.

EDI is worth a paragraph here, because it proves the point by omission. Machine-to-machine exchange between firms has existed for decades, and its setup cost and per-partner mapping have always confined it to the largest trading partners, leaving everyone below that threshold — which is most of the American supplier base — on email and spreadsheets. The control point is that the layer stays external, neutral, and interoperable. Build it any other way and you have shipped another internal silo with a new logo.

Operational Test

The gap is measurable, and I would rather argue about the measurement than the theory. Track the share of inbound RFQs that arrive as unstructured email or PDF versus structured requests; if the enterprise stack had solved coordination, that share would be low, and in the shops I have looked at it is not. Measure supplier onboarding and qualification time for a first-time counterparty. Measure the manual re-entry rate across system boundaries, meaning how many times a human retypes the same number on its way from a buyer's head to a shop's traveler. Count the new suppliers added outside the approved vendor list each year, because that number tells you how much sourcing is happening off the books of every system you own.

A network that is actually coordinating drives onboarding time down and structured intake up. An internal system leaves both exactly where it found them.

Implications

If the enterprise stack could coordinate across firms, the critic would be right and this series would be a long description of a solved problem. It cannot, because it was never scoped to, and the country's coordination failure lives precisely in the seams those systems do not reach. Decades of internal digitization made individual firms efficient while the connections between them stayed analog — and a national industrial base is those connections.

The practical failure mode is mis-scoped tooling: the belief that internal software will eventually stretch to cover an external problem it was never built to touch. It has had thirty years to stretch.

The next objection grants that the layer might be real and doubts anyone will use it. Buyers, the skeptic says, will not change their behavior.

Questions to Ask

1. What share of our inbound work arrives as structured data versus an email attachment?
2. Which of our systems can send a quote-ready package to a supplier we have never worked with?
3. How long does it take us to qualify a first-time supplier, and what does the internal stack do to shorten it?
4. Where does supplier discovery actually happen today, and is it in any system at all?
5. When we add a supplier outside the approved list, what record travels with them?
6. Which coordination work are we doing in inboxes because no system owns it?