Not long ago, a sixty second animated spot with a custom voice, custom art direction and a real edit meant a real production budget and a real timeline. Last month we produced one for about six dollars.
That is not a brag. It is the whole problem, stated plainly.
The floor fell out of production
We built the pipeline behind Hoowee, our own consumer app, where someone records a spoken tall tale and gets an animated short back. Getting the cost down to roughly six dollars a video took real engineering: model choices, a lot of failed passes, and a pipeline that knows what to do when a generation comes back wrong.
The same shift shows up everywhere we work. We run a batch enhancement pipeline for architectural renderings that turns a rough massing study into something a broker can put in front of capital, at a fraction of what a visualization house charges. We automated an InDesign template system with ExtendScript so a brand identity system rebuilds its own collateral instead of a production designer rebuilding it by hand, page by page.
Every one of those was expensive to make and is now nearly free to run. Multiply that across the industry and you get the anxiety currently running through every agency and every marketing department: if making things is cheap, what exactly is anyone paying for?
The ceiling did not move
Here is what the cost collapse did not touch. It did not tell us what the Hoowee spot should be about, or why the tall tale format works and a straightforward story does not. It did not decide which of forty renderings to show a lender, or which one quietly undermines the pitch. It did not know that a brand’s collateral system had the wrong hierarchy in it, which meant automating it perfectly would have automated the mistake at scale.
Cheap production makes judgment more valuable, not less, for a boring structural reason: when output is scarce, the constraint is capacity, and you pay for capacity. When output is abundant, the constraint is knowing which output is worth having, and you pay for that instead. The bottleneck moved. It did not disappear.
You can watch this happen in real time on any social feed right now. The volume of competent, on brand, technically clean content has gone vertical, and almost none of it is memorable. Nobody’s problem is that they cannot make enough. Their problem is that everything they make looks like everything else.
What this actually changes about the work
Three things, in our experience.
Bad ideas get expensive faster. When a campaign took six weeks to produce, a weak concept died in review. Now it ships in an afternoon and you find out it was weak from the market. The review gate matters more, not less.
Volume stops being a strategy. If everyone can post daily, posting daily is table stakes and not a position. What you say has to be worth the slot.
The person holding the tool needs to have done the job. These systems do not fail loudly. They produce something plausible, always, and you need enough pattern recognition to know when plausible is wrong. That comes from having made the thing the hard way for a long time first.
Where we landed
We are not selling AI. We build with it, constantly, because it lets a small shop move at a size it has no business moving at. It is how one operator with a bench of specialists can take on work that used to need a floor of people.
But the deliverable was never the video, the rendering, or the deck. It was the decision about what to make. That part has not gotten cheaper, and we do not expect it to.
