Industrialist Paper No. 24

Interoperability Over Lock-In

Why This Matters

An estimator receives an RFQ for a complex titanium aerospace bracket. The buyer attaches a STEP file exported with full semantic PMI (datums, position tolerances, and feature associations) all intended to be machine-readable. Yet when the file lands in the shop’s CAD and quoting tools, much of that intelligence evaporates. Semantic GD&T often degrades into simple graphical annotations or disappears entirely, especially if the exporter used AP203 instead of AP242, or if the receiving system’s importer lacks full support for associative PMI. The estimator must now manually re-enter critical tolerances, chase the buyer for clarification on intent, or make assumptions that raise scrap risk. What should have been a clean digital handoff of engineering intent becomes days of duplicated effort and elevated cost.

US manufacturing can be better, cheaper, and faster when coordination is fixed. This paper argues for composability over lock-in: clean, documented APIs, structured data portability, and native integration with existing quoting tools, shop management systems, and buyer procurement flows. The goal is a network that plugs into the ecosystem rather than replacing it.

The Persistent Cost of Lock-In

For decades, monolithic ERP and PLM systems dominated by enclosure. Supplier lists, complex BOMs with hierarchical routing, quoting logic, and verification artifacts were pulled inside proprietary schemas. Switching became painful: data migrations involving non-standard formats, duplicate records, and custom mappings routinely drive 55–75% of ERP implementations to miss objectives, with discrete manufacturing seeing failure rates near 73% and cost overruns exceeding 200%. Even successful migrations carry high risk of inaccurate part numbers or incomplete cert data that can halt production lines.

As Paper 23 established, the coordination layer must sit outside the ERP. But it cannot become the next captive platform; not another walled garden that traps participants in proprietary data models and upgrade cycles.

The Irony of the Software Abundance Era

Software creation has never been cheaper or faster. Yet at the moment the U.S. manufacturing base most needs collaboration to maximize domestic capacity, we see the opposite: a rush to build full vertical stacks. Startups and even some shops replicate quoting engines, scheduling logic, supplier databases, and verification modules in-house, promising “end-to-end control.” The result is more fragmentation, not less.

Nowhere is this more evident than the steady stream of “new” US Manufacturing Directories announced on X and LinkedI each promising to solve discoverability while delivering another static list that cannot push structured quotes or capacity data into any buyer’s procurement system or shop’s MES.

Buyers re-key pricing. Shops rebuild capability profiles. Change orders fracture across disconnected tools. Invisible capable capacity sits idle while legacy silos and fresh walled gardens deepen the translation breakdown.

Lessons from API-First Thinking

Early in my time working in Amazon Advertising, I made the case for prioritizing a public, well-documented REST API over closed internal tools. The push faced understandable concerns about security, competitive exposure, and control. Yet opening the API ultimately enabled a thriving ecosystem of partners building custom dashboards, bid optimizers, and reporting layers on top of Amazon’s signals. Development velocity increased, customer adoption accelerated, and the platform grew stronger precisely because it was designed for integration rather than enclosure. The same principle applies here.

Composability in Practice

The coordination layer must function as a neutral network protocol. It exposes versioned, documented APIs (REST or equivalent) for inbound RFQ payloads and outbound structured quotes. It supports portable, schema-defined data formats such as JSON payloads carrying pricing breakdowns, machine capabilities (e.g., axis count, tolerance classes, material compatibilities), PPAP-level verification artifacts, and material cert references that move without re-entry or custom ETL projects.

Buyers retain their familiar procurement portals. Shops keep their existing quoting tools and shop-floor execution systems. The network sits in the middle as translator and router: ingesting heterogeneous inputs (including imperfect STEP files with varying PMI fidelity), applying verification rules, and pushing clean, testable outputs. When full automation is unavailable, graceful fallbacks (structured email templates or human-reviewed portals) keep the chain intact rather than breaking it.

This approach aligns with proven Industry 4.0 patterns: open standards like OPC UA for rich, semantic shop-floor data and MQTT for lightweight messaging enable real-time flows, while well-defined APIs reduce integration costs and unlock predictive maintenance, better OEE, and faster partner onboarding.

Implications

Real-time data exchange across quoting, scheduling, and ERP systems eliminates duplicated effort and accelerates decision cycles. New domestic shops onboard in days instead of quarters. Change orders propagate automatically. The ecosystem gains genuine agility: participants adopt superior point tools without rip-and-replace projects. Coordination becomes computable… capabilities and constraints explicitly routable, integrations observable and testable.

As the series has shown from protocols that create markets (Paper 15) to verification as industrial plumbing (Paper 16) to the coordination layer operating outside the ERP (Paper 23) we rebuild American manufacturing by turning soft failure modes into hard, executable control points that generalists and builders can rely on.

Questions to Ask

• When a new shop responds to an RFQ, how much of the pricing, capability, and verification data must still be manually re-entered or mapped?
• What data formats and schemas are we implicitly forcing on suppliers, and where do those assumptions create latency or errors?
• If we identify a capable domestic partner tomorrow, how many custom integrations or middleware layers stand between their capacity data and our procurement workflow?
• In our current processes, is information flowing as structured, testable signals — or as noise that requires human reconciliation?
• When a design change or schedule shift occurs, does the update propagate automatically across quoting, scheduling, and ERP systems, or does it restart as PDFs and phone calls?

This is what a coordination layer looks like in practice: a neutral translator and router that makes diverse systems interoperate without demanding convergence on any single platform.

The platform that captures the coordination layer eventually captures fragility. The network that remains open compounds national capability. The practical failure mode is continued misallocation; capable domestic capacity staying invisible while we need every available ounce for sovereignty and shock resilience. The open, composable path accelerates the throughput and self-sufficiency we require.