Fall 2023. Rachel — VP of marketing at a B2B software company, sharp, ambitious, the kind of person who reads the room and moves — bought four AI tools in one afternoon.

ChatGPT Enterprise for content. Jasper for social. An ABM platform with "AI-powered intent signals." A conversational tool for the website. The card barely cooled.

She was, briefly, the most forward-thinking person in her company.

For about ninety days.

Because on day ninety-one, the tools — magnificent as they are — needed to talk to the things that actually run the business. The CRM. The automation platform. The data warehouse that three people understand, one of whom already quit.

So Rachel sent a request to IT.

And she met Marcus.

Marcus runs infrastructure. Eleven years in. He has personally watched two "quick pilots" curdle into eighteen-month cleanup jobs, and he has the faintly haunted look of a man who knows where the bodies are because he helped bury a couple of them.

He read Rachel's request — four unfamiliar tools, all of them wanting customer data, none run past him before the contracts were signed — and wrote back two words.

"Not yet."

If you've never worked inside a big company, you might read "not yet" as a timeline. It is not a timeline. "Not yet" is enterprise for absolutely not, not like this, not without a security review and roughly forty-seven forms, and I would dearly love to know who authorized plugging an unvetted AI tool into a database full of personal data we are legally on the hook for.

Rachel read "not yet" as IT being IT. The department of no. The place good ideas go to fill out paperwork and quietly expire.

Both of them were right. That's the part nobody tells you.

And here's the thing. This is not a story about one unlucky company. This is the default setting.

Gartner found that seventy-three percent of AI pilots never make it into production. Seventy-three. And when you go looking for the corpses, almost none of them died of technical causes. The tech worked. The model worked. What failed was two departments who wanted the identical thing and could not, for the life of them, say it to each other in a shared language.

I know exactly what that standoff looks like. I've been the man they put in the room to break it.

A while back I sat down with the marketing and IT leaders of a $122 billion global industrial company — the kind of place where one product line is bigger than most companies you've heard of. Their CMO and their CIO had been circling the same AI problem for fourteen months. Fourteen months. A $122 billion enterprise, frozen, because two genuinely smart executives had each quietly decided the other one was the obstacle.

I'll be honest with you. The first time I ran a session like this, years earlier, I nearly lost the room in the first twenty minutes — because I let both sides start litigating old grievances. A grievance session feels productive and changes nothing. I learned. You don't let them argue about the past. You make them stand in front of the same map of the future until they realize they've been describing the same place in two different dialects.

At the industrial company, that took about five hours. By the end of the afternoon the CMO and the CIO — who'd walked in not making eye contact — were arguing about which use case to ship first. Not whether. Which.

The first one went live in 39 days. Three were in production inside ninety. The thing that had been stuck for fourteen months moved in one afternoon and a follow-through.

Now. The consultants have noticed this problem too, of course. And their fix, delivered in a forty-slide deck for a frankly heroic fee, is that Marketing and IT should "collaborate more" and "build a center of excellence" and "foster a culture of alignment."

I want to choose my words here.

That advice is garbage.

Not wrong, exactly. Just useless — useless the way a sign that reads "be happier" is useless. Collaborate more. Thank you. Rachel and Marcus would adore collaborating more. They can't, because they're using the same words to mean opposite things, and you cannot culture-foster your way out of a translation problem. You don't need a center of excellence. You need a Rosetta Stone and somebody who can read it.

Because Marketing speaks one language. Speed. Pipeline. The competitor who just shipped. Get it live, learn, adjust.

And IT speaks another. Stability. Governance. What happens at two in the morning when the thing breaks. They don't measure speed to market. They measure, more or less, mean time to disaster.

Neither language is wrong. They are simply different languages — and right now your AI strategy is stranded between them like a tourist who memorized three phrases from a guidebook and is now profoundly lost.

Which is the whole reason we built the workshop.

It's called Translating AI, and it does one thing. It puts your marketing leaders and your IT leaders in one room for an afternoon and turns the standoff into a roadmap they both signed.

Half a day. Both teams. We surface the real tensions — not the diplomatic ones, the actual ones from the pre-work, where Marketing wrote "IT says no to everything" and IT wrote "Marketing buys tools and expects us to clean up after." We put your honest AI maturity up on the wall where no one can argue with it. We take the twenty-odd use cases everyone's been quietly fighting over and cut them to the three worth doing first — the ones that move money and won't get anyone fired. And you leave with the one thing most companies don't produce in six months of meetings: a single plan both departments are bought into, with names and dates on it.

I've sat in both chairs. Twenty-five years on the marketing side, and enough time across the table from the CIO to speak IT without a guidebook. I don't show up to tell IT to move faster or Marketing to be more careful. I show up to translate — so the two of them can finally build the thing they both actually want.

Here's how it works. Reply to this and we'll put a date on the calendar. Two weeks out, a fifteen-minute questionnaire goes to each side — that's where the real grievances surface, in private, before anyone's in the room. I read every answer before I walk in. Then we spend one afternoon, and forty-eight hours later you have the roadmap, the priorities, and the governance, in writing.

Because here's the other ending. The one where nobody translates.

Rachel's four tools sit in a drawer. The pilot dies in committee — which is to say it dies in a sequence of meetings so polite no one can name the day it happened. Marcus goes back to keeping the lights on. And eighteen months later a competitor — one whose marketing and IT leaders spent a single afternoon in a room together — ships the exact thing Rachel pitched. Her CEO walks into her office holding a printout, asking why they're behind.

She'll have an answer. It just won't be a good one.

You can be the seventy-three percent. Or you can spend an afternoon and not be.

Get them in a room.

Chris Bannon at Buzzhoney runs the Translating AI workshop — a half-day session that gets Marketing and IT aligned on one AI roadmap both sides can support. Find out more here...

P.S. The pilot that dies in committee never dies dramatically. There's no funeral. One day it's a priority; six weeks later nobody mentions it, and everyone quietly agrees it was probably always going to be hard. That $122 billion company believed that too — for fourteen months. Then they spent one afternoon in a room and shipped in nineteen days. The gap between those two outcomes was never the technology. It was whether anyone got the two of them speaking the same language before the competition did.

‍