The Forward Deployed Creative.

Palantir built its growth around a role called the Forward Deployed Engineer. Instead of selling software and hoping customers figure it out, they send their own engineers to live inside the customer's building until the thing actually works. Not a consultant who hands over a slide deck and leaves. Someone who stays until the product does the job.

Someone pointed me to a piece predicting the same model is about to land in Hollywood. AI video companies embedding creative engineers directly inside post-houses, VFX studios, and ad agencies, building custom fine-tunes and control interfaces that no off-the-shelf tool could match. They're calling it the Forward Deployed Creative playbook.

That's the one prediction in there worth sitting with. Everything else is noise by comparison.

The model obsession is already outdated

Everyone's still arguing about which video model wins this quarter. I get why. It was a fun argument for a while. It's also increasingly beside the point.

Access to a good model stopped being the bottleneck a while ago. AI tools are everywhere in creative work now. Junior editors and department heads are both reaching for them daily. What's still hard is getting a studio's actual footage out the door on brand, on deadline, consistent with years of established visual language. That's not a model problem. That's a judgment problem.

This is the gap a Forward Deployed Creative fills. Not a vendor selling a subscription from the outside. Someone inside the building who knows the characters, the color science, the client's taste well enough to build something custom. Part prompt engineer, part editor, part creative director. Nobody wrote that job description five years ago. Somebody's hiring for it now.

Where the money is actually going

I don't need a research report to tell me creatives who've figured out how to pair taste with these tools are charging more than the ones who just learned to type good prompts. The data backs it up if you want it, but the logic was always obvious. A prompt is replicable. Judgment about a specific brand, built up by sitting inside its world for months, is not.

The real split happening right now isn't AI creatives versus non-AI creatives. It's generalists who use the tools versus specialists who've become fluent in both the technical stack and the craft. One of those groups is interchangeable. The other isn't, and it's getting paid like it.

This is the exact tension we built Popcorn around

Cheap generation was never the hard part. Anyone can produce a clip now. The hard part is knowing what's worth making, keeping it consistent with a brand's voice month after month, and having a real person decide what actually ships. That's why Popcorn isn't a prompt box.

It learns a brand's business and finds the stories worth telling. It turns existing content and IP into new concepts, writes the scripts, produces the video and motion graphics, edits it through a full timeline, and publishes it. Then it uses performance data to decide what gets made next. At every step, nothing goes out without the customer's permission. The AI does the heavy lifting. The human stays in charge.

This is a Forward Deployed Creative function, just built into the product instead of hired one desk at a time. The gap between "a model can generate this" and "this is actually worth publishing" doesn't close because the model got better. It just moves somewhere. Either a studio pays a person to stand in that gap, or it works with a system built to stand in it from day one.

If you're placing bets on where this goes next, stop watching the model leaderboard. Watch who's getting hired to sit between the prompt and the production floor. That's where the value actually is.