How AI is changing RevOps and GTM

March 18th. That's the date everything changed for me.

Our CTO sent a company-wide email. Subject line: New AI Tooling Coming Your Way. In it, a brief description of Claude 101 (here's how you use it) and Claude 102 (want to get fancy? Download the desktop app). And just like that, every single person in our company had unfettered access to Claude Enterprise.

No rollout plan. No training program. No governance framework. Just... here you go.

I've been doing this for 25 years. In that time, I've never once rolled out a tool, a process, a program, or a system without some kind of plan behind it. I've never turned on Salesforce and told people to go figure it out. But that's essentially what happened here.

At first, I was excited. I'd been pushing for Claude access for a while, wanting to hook it into Clay and do all kinds of things I'd been dreaming about. But that excitement faded pretty quickly when I realized what "everyone has access" actually means in practice. Because I'm a control freak. And if you're in RevOps, you probably are too.

You want tools rolled out a certain way. You want processes. You want systems. You want people using the right tools in the right way, not going rogue with something that isn't part of your SSO, that nobody approved, that's creating data in places you don't even know exist.

So yeah, the anxiety set in fast.

The chaos arrived within three days

By day three, someone had set up an auto-responder using Claude. An internal email went out that was, let's say, less than diplomatic. Something along the lines of declining a meeting because there was no agenda. Blunt. Rude, even. And when the recipient pushed back, the sender said, "Oh, Claude did it."

IT scrambled to add some governance. Read/write access suddenly felt like a much bigger deal than it had 72 hours earlier.

That story is funny in hindsight. But it also captures exactly what's at stake. And it's playing out in companies everywhere right now.

Here's what I keep coming back to: companies are about to scale the wrong thing.

When every go-to-market person has an AI teammate, what's keeping your company operating as one company? Because right now, without the right architecture and governance in place, you're not scaling productivity. You're scaling silos. You're scaling inconsistency. You're scaling chaos, just faster.

People are excited. They're diving in. They're building their own little workflows, their own mini automation setups, their own data dumps to SharePoint folders nobody else can find. And they think they're being more efficient. But what they're actually doing is creating a version of my teenager's closet, where everything looks contained from the outside until you open the door and it all falls out on you.

The signals problem underneath the AI problem

Before we get into how to fix this, I want to name something that I don't think gets enough attention.

Companies don't actually have an AI problem. They have a signals problem. They have a data connectivity problem. They have missing data, poor hygiene, inconsistent definitions, and fields that only the admin truly understands because the institutional knowledge lives in someone's head rather than in documentation.

I ran a LinkedIn survey asking people what they were most worried about when it came to AI rollouts. Bad data came up. Governance came up. Productivity concerns came up. You know what almost nobody mentioned? Customer experience.

That bothered me. Because when you think about what's actually at risk here, the emails that could go out to customers, the inconsistent messaging, the gaps in context that AI fills in with something plausible but wrong, the CX implications are real. And they should be front of mind.

Every process, every customer journey map, every piece of tech sits on top of your revenue engine. If the foundation has cracks, AI doesn't fill them. It just moves faster over them.