Nobody is coming to your film

I made a seven-minute film. I knew, before I finished it, that almost nobody would watch it. Not because it was bad — because there is nowhere for it to go.

I think most people making AI films know this feeling and don't say it out loud. You spend three weeks on something. You generate, you cut, you fix the shot where the hands go wrong, you score it, you export it. And then you hit upload, and the counter says 41, and 31 of those are you checking whether the counter moved.

The reflex is to blame the work. The film was too slow. The story was thin. The faces still look like faces from a dream. Sometimes that's true. But it isn't the main thing, and treating it as the main thing sends you back to make a better film for the same empty room.

The distribution layer doesn't exist yet

Every art form gets two things: a way to make it and a way for it to reach people. AI filmmaking got the first one, violently fast, in about two years. The second one hasn't been built.

Look at where an AI film can actually go today.

YouTube wants watch time

YouTube's recommendation system is optimised for one thing: how long you keep people on YouTube. A seven-minute film from an account with no history and no watch-time record is the worst possible bet for that system, so it doesn't make the bet. You are competing for the same slot as a podcast clip that holds people for forty minutes. Your film can be the better piece of work and still lose that competition every time, because they aren't being judged on the same axis.

TikTok and Reels want thirty seconds

Short-form platforms are extraordinary at distribution — for short form. Cut your film down to a thirty-second hook and it can travel. But the thing that travels isn't your film. It's a trailer for a film nobody is going to go and watch, because there is no obvious "go and watch it" to click through to. You get views. You don't get an audience.

Festivals don't have a category for you

Some are starting to. Most still route AI work into an "experimental" or "new media" sidebar, or exclude it in the rules. And festivals were never a distribution layer anyway — they're a validation layer, with a submission fee, and a few hundred people in a room.

The AI communities are about the tools

The Discords and subreddits and X threads are alive and generous and genuinely useful. But go look at what gets attention there: workflows, prompts, model comparisons, "how did you get that camera move". They're a guild, not an audience. People show up to learn how you made it. Almost nobody shows up to watch it.

So four doors, and none of them open onto people who came to watch a film.

What actually helps

I don't have a clean answer. I have four things that are more useful than making another film into the void.

Be findable, not just uploaded

Uploading is not publishing. A film with a title, a real description, the tools you used, the year, a proper thumbnail and a page that a search engine can read is a thing that can be found six months from now. A file behind an autoplay feed is not. Most AI films are effectively invisible the day after they're posted, and that's a metadata problem more than an art problem.

Make the arrival worth it

When someone does land on your film — from a comment, a link, a friend — what do they find? If it's a bare player on a grey page, they watch forty seconds and leave. If they find the film, a real description, what you used, what you were trying to do, and three other things you made, some of them stay. The difference between a view and a viewer happens in that first ten seconds of context.

Show the making, not just the made

This is uncomfortable if you want the work to stand alone. It also happens to be where the attention currently is. The process is legible and shareable in a way the finished film isn't — and the people who follow you for the process are the people who show up for the next film. Breakdowns, failures, the shot you regenerated fourteen times. The failures travel better than the successes.

Find fifty people, not a million

A million views is a lottery ticket. Fifty people who know your name and open your next thing is a career floor. Fifty is reachable by hand — by answering people, by showing up in the places where your kind of work is discussed, by being someone specific rather than a channel. It's slower and it's less fun to say out loud, and it's the only part of this I've seen actually compound.

Why I'm building SHOXA

I should be straight about this, because you're reading it on the site.

I'm building SHOXA because after making that film I went looking for the place I'm describing — a site where AI films sit as films, with a proper page, credits, the tools, the creator behind them, and a viewer who came to watch rather than to scroll — and it didn't exist. So this is the attempt at it: a place where an AI film gets treated like a film instead of like content.

Here's the part I'd rather not write. SHOXA does not have an audience yet either. If you upload today, you are not going to get a hundred thousand views. What you get is a real page for your film that is built to be found, a profile that makes you look like a filmmaker instead of an account, and a platform whose entire incentive is to send people toward films — because we don't have a feed of forty-minute podcasts to send them to instead.

That's a smaller promise than the one you usually get. It's the one I can actually keep.

If you've made something and it's sitting at forty views, put it somewhere it can at least be found. Then keep going. The work isn't the problem.