We've Spent 20 Years Renaming Operations. The AI Era Will Force Us to Actually Redesign It.

Here is the specific thing that's been nagging at me.

Most modern GTM teams have more data than they've ever had. More platforms. More signals. More dashboards. More automation running in the background twenty-four hours a day. And yet the quality of decisions that come out of these systems often hasn't materially improved.

We've spent billions on data infrastructure and come out the other side with more charts and more opinions about which chart is right.

The industry has no shortage of Ops labels. What it lacks is a shared understanding of what those labels should actually produce — not reports, not dashboards, not workflows, but confident commercial decisions that the business can act on.

Most companies have data abundance and signal scarcity. Marketing knows how many contacts opened an email. Sales knows which accounts viewed the pricing page. CS knows which customers logged in twice last month. But no one is systematically turning those raw inputs into decision-grade signal — clean, contextualised, actionable reads on what the business should actually do next.

Instead, what happens? Each silo holds its version of the truth. Forecasting meetings become negotiations between conflicting datasets. Routing rules are tribal knowledge maintained in someone's head. The handoff from Marketing to Sales is still, in most organisations, a handoff between two spreadsheets dressed up as an integration.

We built more systems of record. We didn't build better systems of judgement.

Beneath the tooling and the taxonomies, what leaders have always really wanted — and rarely been able to get consistently — is confidence in their commercial decisions.

If this were just an operational efficiency problem, it would be annoying but manageable. But it isn't.

AI is about to make this very expensive to ignore.

The reason is not the one most people are focused on. It's not about AI replacing headcount, or which tools will survive the platform consolidation, or whether your RevOps analyst will be replaced by an agent. Those are real questions, but they're not the most important one.

The more important question is this: AI is fundamentally a decision-amplification technology. Every AI capability in your GTM stack — predictive scoring, automated routing, intelligent nurture, AI-assisted forecasting — is making a decision on behalf of the business. Routing a lead. Flagging a churn risk. Personalising a message. Recommending a next best action. And doing this at a speed and scale that no team can manually review.

So here's the problem. If the decision logic embedded in those AI systems is built on fragmented, conflicting, silo-maintained data, the AI doesn't fix it. It executes it. Faster. At scale.

This is not a warning against AI adoption. It's an observation about foundations. If your GTM architecture was producing mediocre decisions at human speed, AI will produce mediocre decisions at machine speed. The stakes go up. The visibility of the problem goes up. The cost of bad decision quality goes up. And unlike a bad hire or a bad campaign, a bad decision model running inside an AI system touches every account, every lead, every intervention — simultaneously.

There's a second effect worth naming explicitly, because I think it reframes what Operations leadership is actually for.

For the last decade, the competitive advantage in GTM Operations was largely about execution capacity. Who could automate more. Who could run campaigns at greater scale. Who could move faster across more channels. Operational excellence meant speed, volume, and throughput.

AI is commoditising that. If your competitive edge is "we can run a lot of outreach" or "our team can process a lot of leads" or "we move faster than competitors" — that gap is closing. Fast.

Execution is becoming abundant. Decision quality is becoming scarce. And scarcity is where value concentrates.

What doesn't commoditise is the quality of what you're executing. Whether your ICP thesis is actually right. Whether your scoring model is identifying the accounts that will close. Whether your churn intervention logic is catching the right signals before it's too late. Whether your forecast is based on pattern recognition or genuine commercial judgement.

This reframes what Operations leaders are being paid for. Not just to build systems that execute efficiently. To build systems that decide well. The future company will run on decisions, not just on reports.

So what does a redesigned operating model actually look like?

Let me describe a model I've been working through — not as a framework to adopt wholesale, but as a way of naming what's missing in most GTM architectures today.

A necessary clarification first. This is not an argument to dissolve or replace Marketing Ops, Sales Ops, CS Ops, or RevOps. Each of those is a specialist domain that requires specialist expertise. RevOps remains important. What I'm challenging is the assumption that the architecture connecting those functions should keep being designed around functional silos as AI becomes embedded in every workflow.

The four layers I think are underdeveloped, or missing entirely, in most organisations:

• Signal Operations — The function whose job is to turn data abundance into decision-grade signal. Not to report on what happened — to synthesise what it means and what should change.

• Decision Operations — The function that owns how decisions get made, not just whether they get executed. Who sets the logic? Who resolves conflicts between competing signals? In most organisations, this is implicit and tribal. Decision Ops makes it explicit.

• Execution Operations — The automated execution of decisions through human and AI workflows — including governing the fleet of agents scoring, routing, and nurturing on behalf of the business at scale.

• Trust Operations — The function that governs the AI-augmented stack: accountability, audit trails, override protocols, risk thresholds, human-in-the-loop design. As AI takes on more decision execution, this becomes non-negotiable.

These four layers aren't a new department. They're a set of design principles and ownership structures that run through — and connect — your existing functional teams. The connective tissue is what I'd call Decision Architecture: the intentional design of how information flows through your GTM system, gets synthesised into signal, gets processed into a decision, gets executed by humans or agents, and gets governed for accountability.

Most companies have the first and third of those steps reasonably well covered. The second and fourth barely exist.

Some of the roles that a decision-architecture model requires will look unfamiliar. That's the point.

The people who run Decision Operations will need to understand both the commercial logic of a business and the technical architecture of AI systems — simultaneously. The people who run Signal Operations will need to think about data synthesis as a product, not just a reporting function. The people who govern AI agent fleets will need to understand operational risk in a way that previous automation roles never required.

None of this makes existing Revenue Operations or Marketing Operations roles obsolete. It makes them more important, and more connected. The specialist who deeply understands the lifecycle of a marketing-qualified lead is still essential. What changes is the system they're operating within, and the additional functions that system now requires.

The title that I think will matter more over the next five years than any of the titles we've been debating is one we haven't formalised yet: Head of Decision Operations. The person who owns how the business makes commercial decisions, not just whether those decisions get executed. And at the executive level: Chief Decision Officer — not a title that replaces the CRO or CMO, but one that owns the quality of commercial judgement as a function of enterprise infrastructure.

The gap between where most organisations are today and where they need to be isn't small. And the transition won't be smooth.

Here's what I think happens over the next five years with companies that don't address this.

AI systems get deployed into GTM stacks that were never designed for AI. The systems are fast and always on, but they're operating on fragmented signal and inconsistent decision logic. Results are uneven. Some things work, many don't. The internal assumption is that the AI tools are the problem — that a better model, a different vendor, a more sophisticated prompt will fix it.

It won't. The architecture was the problem before the AI arrived. The AI just made the consequences more visible, more frequent, and more expensive to reverse.

The organisations that start building decision architecture in the next 18 months will have a compounding structural advantage over the ones that don't. And the ones that don't will spend the next decade doing what we spent the last two decades doing.

Renaming the problem.