The Future ECO Process: From Approval Workflows to Collaborative Change Exploration

Engineering change is quietly outgrowing the systems we built to control it.

On paper, ECO and ECN processes still look reassuringly familiar. They define steps, route approvals, manage revisions, and create the sense that change is contained and governed. Every PLM system has its version of this logic, and most organizations depend on it to keep order.

But when I spend time with engineering and manufacturing teams, I see something very different happening in practice.

The real work of change rarely starts with an ECO form. It starts with uncertainty, discussion, and negotiation—long before anything is officially approved. And as teams become more distributed and products more complex, the gap between formal change processes and real change work is becoming impossible to ignore.

This realization connects directly to a question I explored in an earlier OpenBOM article:

👉 What does an AI-enabled collaborative workflow actually look like?

It also aligns with a broader industry discussion about agentic workflows and context graphs, including Foundation Capital’s perspective on why decision traces, not just data, but represent a massive opportunity for AI-driven systems: AI’s trillion-dollar opportunity: Context graphs.

When these ideas come together, they point toward a different future for engineering change—one that moves beyond approval workflows and toward collaborative change exploration.

Where Change Actually Begins

Let’s start with reality, not process diagrams.

Engineering change almost never begins with a clean problem statement and a ready-to-route ECO. It begins with something not working as expected. A supplier misses a delivery. A part suddenly goes on allocation. A tolerance turns out to be unrealistic in production. A cost target is exceeded. A test result exposes an edge case no one anticipated.

The first response is not administrative—it’s investigative.

Engineers ask whether an alternate is possible. Procurement looks at availability, lead times, and pricing. Manufacturing evaluates whether the change introduces risk or rework. Quality raises compliance questions. Sometimes suppliers are pulled into the conversation to clarify what is feasible and what is not.

These conversations unfold across meetings, emails, spreadsheets, shared documents, and chat threads. They often span days or weeks and cross organizational and geographic boundaries.

Only after most of this thinking is done does an ECO appear.

By that point, the ECO is not guiding the decision—it is documenting the outcome. The rationale, tradeoffs, and exceptions that shaped the decision remain scattered across tools or lost entirely.

This is where traceability breaks down. Downstream teams receive an approved change but not the reasoning behind it. As a result, they hesitate to trust it fully. They double-check. They ask the same questions again. Manual verification becomes a safety mechanism rather than an exception.

This is not a failure of discipline. It’s a mismatch between how change actually happens and how our systems expect it to happen.

Why AI Cannot Fix This Alone

Given this backdrop, it’s tempting to assume that AI will simply “fix” change management by analyzing data faster or automating impact analysis.

But AI cannot compensate for missing context.

If the system only contains finalized revisions and approval stamps, AI has no visibility into intent. It cannot tell whether a change was a deliberate strategic compromise, a temporary workaround, or a risky exception accepted under pressure.

This is why many AI-assisted change initiatives feel underwhelming today. They operate on thin data—snapshots without product memory.

In earlier discussions about agentic workflows, one idea stood out clearly: AI systems that operate over long time horizons require persistent, structured context. They need to understand not just what changed, but how and why decisions evolved over time.

That point is emphasized in this talk on long-running agentic processes by one of my favorite AI researchers – Nate B. Jones:

AI does not become valuable because it is “smart.” It becomes valuable because it has access to workflow memory—the accumulated reasoning embedded in day-to-day work.

Engineering change is one of the most important places where that memory should exist.

Reframing ECO as a Shared Decision Environment

This leads to a necessary reframing.

ECO is not fundamentally an approval workflow. Approval is just the visible endpoint.

At its core, ECO is a decision-making process that unfolds over time, involves multiple stakeholders, and balances competing constraints. Treating it purely as a routing problem strips away its most important characteristics.

What teams actually need is a shared decision environment—a place where uncertainty can be explored openly, alternatives can be evaluated in context, and reasoning can travel forward with the product.

This is what I mean by a collaborative workspace. In a collaborative workspace, product data is not isolated from discussion. Comments are attached directly to items, assemblies, and BOMs. Tasks represent open questions and unresolved risks, not just to-do lists. Multiple people can work on the same data at the same time, seeing each other’s input rather than waiting for handoffs.

Most importantly, decisions are not frozen into forms prematurely. They mature through collaboration, and when they are finally approved, the reasoning behind them remains accessible.

This shift, from routing approvals to supporting shared understanding, is foundational for any meaningful evolution of the ECO process.

Decision Traces and the Role of Context Graphs

The Foundation Capital article on context graphs puts words to something many teams feel intuitively.

Organizations don’t really operate on static objects like parts and documents. They operate on decisions. And decisions leave traces – exceptions, overrides, approvals, rejections, and precedents.

When these traces are captured and connected, they form a context graph: a living structure that reflects how the organization actually reasons.

Engineering change is a natural generator of decision traces. Every ECO answers implicit questions about risk tolerance, priorities, and acceptable tradeoffs. Every exception encodes organizational judgment.

When ECO systems only record final states, those traces disappear. But when discussions, tasks, and approvals are linked directly to product data in a graph-based model, context emerges naturally.

This is why data modeling matters. A graph-based representation makes it possible to connect items to decisions, decisions to discussions, and discussions to outcomes—without forcing everything into rigid workflows.

An ECO Scenario, Revisited

Consider a supplier-driven change involving a critical component that suddenly becomes unavailable.

In a traditional setup, engineering looks for alternates in a spreadsheet. Procurement negotiates via email. Manufacturing waits for clarity. Eventually, an ECO summarizes the conclusion, but not the path.

In a collaborative workspace, the path is the process.

An engineer proposes an alternate directly on the affected item. Procurement adds context about cost and MOQ. Manufacturing creates a task highlighting a potential assembly issue. Quality asks a question about compliance.

These interactions are not peripheral—they are the change process itself.

When approval finally happens, it doesn’t erase the discussion. It closes it.

At that point, AI can play a useful but restrained role. It can summarize discussions for stakeholders who join late. It can automatically identify impacted assemblies. It can surface similar past decisions and their outcomes. It can suggest next steps based on patterns, not rules.

AI doesn’t replace judgment. It supports product memory.

What Changes in Practice

When change management moves into a collaborative workspace, the effects are subtle but powerful.

Teams spend less time manually rechecking decisions because the reasoning is visible. Downstream groups trust change data more because they can see how decisions were made. Change cycles shorten because fewer questions need to be rediscovered and revalidated.

Over time, something even more important happens. The organization begins to accumulate product memory—not in documents, but in connected experiences. Each change becomes a reference point for future decisions.

This is the real foundation for AI-assisted change management.

From Concept to Practice

This is not a future vision waiting on technology.

As part of the OpenBOM Tasks and Comments beta, customers can already start working this way—using a collaborative workspace to explore changes, assign tasks, and capture reasoning directly in context. 

Check our earlier article – OpenBOM Review: BOM Collaboration with Comments and Tasks and the following demo video that will give you an idea. 

Also check OpenBOM collaborative workspace article explaining how simultaneous editing is used to prepare a change process – OpenBOM collaborative workspace technology in practice. 

For teams interested in shaping what the next generation of ECO processes should look like, this is an opportunity to participate as design partners—to experiment with collaborative, context-driven change workflows today and influence how they evolve.

Because the future of ECO is not about better forms or faster approvals.

It’s about building environments where decisions are made collaboratively, remembered reliably, and reused intelligently.

I’m curious: How does your team handle change decisions today—and what do you wish your ECO system actually remembered when someone asks, months later, “Why did we do it this way?”

REGISTER FOR FREE to check how OpenBOM can help you and also discuss a design partnership with OpenBOM to check new AI-enabled collaborative ECO workflows. 

Best, Oleg