AI Video Is Here. What It Actually Changes for Small Teams

The past month has been loud. Generated video went from a research demo to something in everyone's pocket, and my feeds are full of clips that would have taken a motion designer a week just two years ago.

I have run small production teams for two decades. Church livestream crews built out of volunteers. A small video production company I founded and ran for a few years. Conference coverage with a team you could count on one hand. I have also spent the past couple of years deep in AI tooling on the workflow side, so I am neither a skeptic with arms crossed nor a hype account. Here is my honest read on what this wave actually changes for small teams, and the two things I am convinced it does not touch at all.

What actually changes

Previsualization is now free. I can storyboard a shoot in an afternoon with generated frames instead of stick figures. Clients understand a moving mock instantly. That alone changes preproduction conversations.

Let me make that concrete, because previsualization sounds like a luxury item and it is not. For most of my career, the gap between "here is what I am imagining" and "here is what you will get" was bridged with words, reference links, and trust. Sometimes a rough storyboard if the budget allowed. Clients would nod along in the pitch meeting and then be surprised at the review, not because the work was bad but because the picture in their head never matched the picture in mine. Every one of those mismatches costs a revision cycle, and on a small team, revision cycles are the budget.

Now I can walk into a preproduction conversation with moving frames that are eighty percent of the visual idea. Wrong in the details, absolutely. Disposable, completely. But the client reacts to something real, and their notes come before the shoot instead of after it. The generated material never ships. It exists to move the misunderstanding to the cheap end of the project. That is worth more than any finished asset these tools will ever produce for me.

Filler footage is commoditized. Generic b-roll, abstract backgrounds, motion textures. The stuff you used to buy from stock libraries or burn an evening making in After Effects is now a prompt. Budgets should shift away from it.

I want to be precise about which footage I mean, because this is where people either overclaim or panic. I am talking about the connective tissue: the slow push across a desk, the abstract particle background behind a sermon title, the drone-style establishing texture that nobody looks at directly. Footage whose entire job is to not be noticed. I have spent real evenings of my life building that material, and I have spent real money licensing it, and I felt no craftsman's grief watching it become a prompt. It was never the work. It was the packing material around the work.

The budget implication is the part small teams should sit with. If a chunk of your line items has historically gone to stock licenses and filler animation, that money is now available for the things that still require humans: another hour of interview time, better audio, an extra camera on the day that matters. The teams that quietly reallocate will look noticeably better than the teams that keep spending the old way.

One person can carry more of the pipeline. A solo communicator at a small organization can now produce things that used to require a bench of specialists. For the small churches and nonprofits I work with, that is not a threat. That is the whole opportunity.

This is the change I care about most, because I have sat in the room with that solo communicator dozens of times. Most of the organizations I have served in Georgia and Alabama never had a media department. They had one person, often part time, often a volunteer, doing graphics and video and social and the website, permanently triaging. For that person, this technology is not replacing a team. There was never a team. It is the first time they can produce previsualization, motion elements, and rough cuts at a quality that used to require specialists they could never have hired.

The honest flip side: the ceiling on "acceptable" rises for everyone at the same time. When decent becomes easy, decent becomes the floor. Which brings me to what does not change.

What does not change

Real people are the point. A testimony, an interview, a pastor speaking to their own congregation, a founder telling you why they started. Nobody wants a synthetic version of that, and audiences can smell it. Capture work involving real humans in real rooms just became more valuable, not less.

Twenty years of church production taught me exactly one unshakeable thing about what moves an audience, and it was never production value. It was a person on camera saying something true. I have watched congregations sit politely through beautifully produced announcement videos and then lean forward, actually lean forward, when someone stood up and told their own story badly lit and slightly out of frame. The power was never in the pixels. It was in the fact that a real person was really saying it.

Generated video cannot produce that, not because the pixels are wrong but because the claim is wrong. A testimony is only worth something because it happened. The moment an audience suspects the human in front of them is synthetic, the value does not degrade, it inverts. And audiences are getting suspicious fast. I already see people reflexively checking whether a clip is real before they let themselves feel anything about it. In that environment, verifiably real footage of real people becomes a premium product. If your team is good at getting humans comfortable on camera, at drawing out the honest sentence instead of the rehearsed one, your stock just went up.

Taste is still the bottleneck. The tools generate. They do not decide. Knowing what the story is, what to cut, what the brand sounds like, and when something is simply not good enough. That was always the hard part, and it still is.

Here is how this plays out practically. Generation is cheap now, so everyone generates a lot. I have watched people produce forty clips in an evening and end up with nothing usable, because the skill was never producing clips. The skill is looking at forty clips and knowing which one serves the story, or that none of them do and you should shoot it for real. The tools have moved the bottleneck, not removed it. Before, the constraint was execution: could you make the thing? Now the constraint is judgment: should this thing exist, and is it right?

Judgment is trained the slow way, by making a lot of work and watching how audiences actually respond to it. Two decades of Sundays gave me thousands of those feedback loops. That is the asset that survives this transition. If anything, the flood of generated content makes trained taste easier to spot, because it is now the only scarce ingredient in the room.

The failure mode I keep seeing

Since everyone asks what could go wrong, here is the pattern I keep running into: small teams aiming the new tools at the wrong end of their pipeline. They generate the centerpiece, the emotional core, the thing the whole piece exists for, and then wonder why the result feels hollow. Meanwhile they are still hand-building the filler that the tools are genuinely good at.

It is exactly backwards. The centerpiece is where the humanity has to live. The filler is where the automation should live. When I consult with a team on this now, the first exercise is boring on purpose: list everything you produced last quarter and sort it into two piles. Pile one is anything where the audience cares that a human made it or that a real moment happened. Pile two is everything else. Automate pile two ruthlessly. Protect pile one completely. Most teams find the split runs somewhere near 40 percent automatable, and almost none of them had their effort allocated that way.

The other failure mode is subtler: letting the tools set the creative agenda. Because generation is fast and fun, teams start making things because the tool can, not because the audience needs it. Output goes up, impact stays flat. Volume was never the goal. If the new capacity does not translate into either better work or recovered hours, you have adopted a toy, not a tool.

What I would do this quarter

My advice to small teams: learn the tools now, aim them at the boring 40 percent of your pipeline, and reinvest the saved hours in the human footage only you can get.

In practice, that looks like three moves. First, run the two-pile exercise above and be honest about it. Second, pick one recurring production, your weekly announcements, your monthly recap, whatever ships on a schedule, and rebuild just its filler layer with generated material. A recurring piece gives you repetitions, and repetitions are how you actually learn a tool's character instead of just its demo reel. Third, take the hours you recover and spend them on capture: schedule the interviews you have been postponing, sit with people, get the stories that only exist in your community.

A year from now, I do not think the winning small teams will be the ones with the best prompts. I think they will be the ones who used the cheap parts of this technology to buy back time, and spent that time pointing cameras at real people. The tools changed what is easy. They did not change what matters. Teams that keep those two lists straight are going to have a very good decade.