When I first moved into tech project management, I made an assumption that shaped how I thought about my work.
Technology makes or breaks projects.
Pick the right tech. Implement it well. The system works. The project succeeds. Simple.
That assumption is only half the story. And the half I was missing is exactly why I’m not worried about AI coming for my job.
A few years ago, I got pulled into a project with a clean-sounding goal—30-day turnaround on recurring services by end of quarter. I joined my first call for the project and listened to who was doing what and why.
Operations was drowning in a backlog and wanted it gone. Product was building new automations—which, fine, except those only applied to services initiated after the features shipped. Nothing for the backlog. Finance was there because revenue only counts when a service is fully delivered, not when it’s sold. And business operations was building dashboards for all three—three dashboards, telling three different stories about what we were solving and how close we were to solving it.
Early in my career, I’d have assumed the disconnect was me. That everyone else could see how the pieces fit and I was the one who couldn’t. I’d have slunk off after the meeting to quietly ask my manager to explain what I’d missed.
By now I knew better. There are always one or two simple questions that tell me whether it’s a me problem—I don’t understand yet—or a project problem—nobody actually knows what we’re trying to do.
So I asked the room: “What does it mean to be done?”
Blank stares.
Fine. Reverse-engineer it. “Let’s take one item from the backlog. How do we know this one isn’t done? And is that answer different for a new incoming service, or the same?”
Six people. Six different answers.
That’s not a me problem. So the highest-priority task on the project became: define done.
I saw it again a while later, different client, when a business case for an AI implementation landed on my desk. I couldn’t make it make sense—but I hadn’t worked with that business line yet, so I gave it the benefit of the doubt and got the leaders and managers in a room. Same question. “How will we know we’re done?”
“When the AI can analyze these spreadsheets.”
Okay. “How will we know it has analyzed them? How will we know it hasn’t?”
About 5 different answers and if/then scenarios. Same panic. Different disco.
Look at what actually happened in those rooms. Nobody was wrong. Operations was right that the backlog was the fire. Product was right that automation was the fix—for the wrong half of the problem. Finance was right that none of it counted if the numbers didn’t line up. Each of them held a real piece. None of them held the whole thing.
That’s always the hardest part of every project. Everyone is a little bit right and a little bit wrong, and whether you can possibly do a good job comes down to one question: can you get them to agree on what success looks like? Not which person wins. What done means. BIG picture. And it looks different on every project—different people, different pieces, a different definition.
So the job isn’t knowing the correct answer and executing it. It’s reconciling a room full of partial answers into one everyone will actually stand behind.
I used to think my subject matter expertise was what made me a great production manager. My years of experience working on reality shows helped me produce better reality shows—it mattered. But I knew it wasn’t everything. I knew that I had a core skillset that would help make me a great project manager in other industries too. But I wasn’t sure how much a lack of subject matter expertise would impact my overall performance. When I was trying to land my first job in tech, that expertise would be gone. No domain to stand on. Enterprise software implementations, data systems—I would have to learn the ground while walking on it. I’d have nothing to lean on outside of my abilities to orchestrate teams of experts and information, so I’d have to lean on that hard and build the technical knowledge in parallel.
And that’s when it clicked: this was exactly what I’d done in television.
A home renovation show—I had to learn how a home reno project runs. A cooking competition—I had to learn how a kitchen runs. I never once succeeded by becoming a real construction PM in six weeks, or a sous chef. I succeeded because I could stand between experts who didn’t speak the same language and get them pointed at the same outcome. I relied on the trades, the chefs, the crew; they relied on me to see the whole board they couldn’t. That was my subject matter expertise. It was never the electrical code. It was never the pantry quantities. It was the vantage point. It was knowing how to have their back and how to show them what it looked like to have mine.
The vantage point is half of it. The other half is what it takes to build one from nothing: walking into a situation with no map. No brief, no precedent, nobody who can fully articulate what they need. There’s no one way to light a scene, no one way to cut one, no one way to run a show that’s never been made before. So you learn to sense what is not being said, just as much as you learn to coach them how to tell you what they are looking for. You learn to extract the requirement from people who can’t quite name it, pressure-test your read until you’re sure you caught what they meant and not just what they said, and then execute. And honestly, the execution is the easy part.
That’s the piece entertainment people never credit themselves for, because everyone around them has it too. On a set, ambiguity is as certain as a call sheet. You only find out it’s rare when you leave—when you’re in a corporate room full of people who built their careers on being handed clean requirements, and you watch them freeze at the exact moment you’d start moving.
That’s the foundation I brought to corporate. And it’s exactly why the AI question doesn’t scare me.
Automation isn’t new to project management. We’ve had it for years, and it didn’t do the thing everyone’s now afraid it will.
Take Jira. Long before AI showed up, it had powerful automations for most of the administrative overhead of work and project management—task tracking, status reporting, timeline visibility, dependencies. On paper, it should have eliminated the busywork of PMs. If automating the mechanical parts were enough to run a project, Jira would have proven it a decade ago.
It didn’t. In almost every organization I’ve worked with, Jira adoption is mediocre at best. People don’t update their tickets. They don’t log time. They forget to close things out. The system becomes friction instead of help.
Not because Jira isn’t capable. It’s incredibly capable. It’s because the software can’t touch the part that actually decides whether a project works—the part from that first meeting. It can’t get a room to agree on what it’s building. Hand it perfect inputs and it’ll produce a flawless record for a project nobody’s aligned on, which is just a faster way to build the wrong thing. Hand it inconsistent inputs and it’ll get out of sync with reality and become another source of misinformation and confusion.
A human fixes that. Someone who can see that the shortcomings of technology aren’t technology problems at all—it’s that nobody connected the way people actually work to the outcome they actually care about. Same gap as the first meeting, wearing different clothes.
AI can automate even more of this. Auto-populate the fields. Generate a work breakdown from good documentation. Make the system easier to use.
And it’ll go further than the old tools ever could. Point it at the fog and it comes back with a requirement—clean, confident, beautifully reasoned. Which sounds like the whole problem solved…in a silo. And if this isn’t the project management tale as old as time, I don’t know what is. Remember the six people who each wanted something different? Six destinations, now with serious momentum. None of them going somewhere that actually solves the problem for the business.
The thing that would have caught that is the thing AI can’t do. It can’t read between the lines. It can’t notice that someone said all the right words to agree in the meeting but kept dancing around a risk that could derail the whole project. It can’t catch two people saying the same words and meaning completely different things.
Which—if you sat in the meeting where six people defined “done” six different ways—is the entire job.
If you’ve built a career in entertainment and you’re eyeing project management, here’s what I want you to know.
You’ve been doing this work your whole career. Walking into ambiguity and making things work anyway—not by becoming the expert, but by standing between people who didn’t speak the same language and getting them to the same outcome. Reading rooms. Building trust across teams that don’t naturally trust each other. Leaning on experts because you were never the authority on everything, and doing it over and over, under unpredictable conditions, on deadlines, with tired people who still had to shoot.
And don’t read this as just a producer thing, or a production-office thing. It’s the whole craft. There’s no one way to light a scene and no one way to cut a show, which means every creative and technical role runs on the same move. Work with an experienced director and she might give you exactly what you need, handed over in your own technical language. Work with a first-time writer/director and you might get a picture locked in someone’s head with no way out—and whether you do a good job comes down entirely to whether you can pull out of him something he doesn’t even know he knows. Gaffer, editor, camera op, PA—it doesn’t matter. You’ve spent your whole career extracting the requirement. You’ve just never had to call it that.
That’s not a niche skill. That’s the skill—the one that decides whether projects succeed.
How you track, communicate, and monitor project work matters. But it’s not what makes you valuable, and it’s the first thing to get automated. So if you’re wondering if project management is still worth pursuing in an AI-enabled world, worry about the right thing. Worry that if all you bring is task tracking and status reporting, you won’t be worth much. Bring the ability to cut through ambiguity and put a shape around it and you’re not just safe from automation—you’re the reason the project works.
You don’t need to learn that. You need to recognize it in yourself and lean in harder.
If you’re considering whether the PMP is right for you—or how to position your entertainment background for project management roles—I’ve written two detailed pieces on exactly that. The first walks through why I recommend it and whether you actually qualify. The second breaks down how to translate your production experience into the language hiring managers understand. Both are available to paid subscribers.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.