What AI actually does (and doesn't do) in our pipeline

Clients ask about this more often now, usually some version of "are you using AI for this." It's a fair question, and it deserves a straight answer rather than a sales pitch either way. So here's where AI genuinely helps in our process, and where we've learned it creates more problems than it solves.

Where it earns its place

We use AI where it speeds up exploration without touching the final craft. Asset generation on 3D projects, quick style exploration before we commit a direction, rapid iteration when we're testing a few visual approaches early on. In these spots, it's a faster sketchbook, not a replacement for the thinking that decides what actually gets built.

That distinction matters. AI is useful for generating options quickly. It's not the thing making the creative decision about which option is right for your brand, your audience, or your brief. That judgment call still sits with our senior team, the same people who've been doing this for over a decade.

Where it creates problems

The clearest place we've seen it fall short is revisions. When a client asks for a change to something that was AI-generated, there's often no clean way to make a targeted edit. You end up regenerating and hoping the new version keeps everything you liked from the last one, which it frequently doesn't. A human-built asset can be adjusted precisely. An AI-generated one often can't be, not without starting over.

We've also found it's not reliable for anything that needs to hold a consistent, specific brand identity across an entire piece. Style drift is a real problem once you're stitching together multiple AI-generated elements into one coherent film.

How this shapes our approach

AI sits in the kit as a tool, not as the strategy. It's genuinely useful for the exploratory, early-stage parts of a project where speed matters more than precision. It's not where we'd put anything that needs frame-accurate control, exact brand consistency, or straightforward revisability once a client has feedback.

If a project seems like a strong fit for AI-assisted exploration, we'll say so. If it isn't, we'll say that too, rather than defaulting to it because it sounds current.

A few practical questions

Will you tell me if AI was used in my project? Yes, if it's relevant to ask about. We're not precious about it either way, since it's one tool among several, not a marketing angle.

Does using AI make projects cheaper? Not meaningfully, in our experience. Where it saves time is early exploration. The animation, design and revision work that makes up most of a project's time still happens the same way it always has.

Are you worried AI will replace what you do? Not for the parts of the job that matter most, judgment, taste, and the ability to revise precisely based on feedback. Those are exactly the things AI still handles worst.

If you're weighing up a studio that leans heavily on AI generation against one that uses it selectively, it's worth asking specifically how revisions get handled. That's usually where the real difference shows up.

‹ Why we require phased sign-off

Why We Quote Off an Animatic, not a Script ›