Less doing. More directing.

There’s a moment I recognize now, even though I didn’t name it at the time. I was reviewing renders at one in the morning — not because the work demanded it, but because the day had disappeared into something else: budgets, emails, chasing invoices, a meeting that could have been a message. The creative work had been pushed to whatever was left over.

That’s what I changed. Not the work. Everything around it.

What I didn’t automate

Before I get into what I did, let me be clear about what I didn’t.

I don’t use generative AI in production. I don’t generate images, I don’t render with diffusion models, I don’t replace CGI with prompts. The visual work that leaves this studio is made by hand, with judgment, with Maya and V-Ray and decisions that took years to develop. That doesn’t get automated — because it shouldn’t. It’s exactly what a client is hiring for.

What I automated is everything that isn’t that.

Pre-production: more iterations, better decisions

The most valuable change wasn’t in production. It was before it.

Pre-production used to be a silent bottleneck. A brief would arrive, you’d read it, ask questions, wait for answers, ask again. The project would take days to take shape before anyone touched a single file. That process is different now.

I can iterate through a concept ten times before presenting it once. I can explore directions I used to discard for lack of time. I can walk into a meeting with three well-argued paths instead of one half-formed idea. Clients make better decisions because I arrive better prepared.

It’s not that the AI has the ideas. It’s that it lets me explore all of them before choosing.

Studio management: reclaiming the time that was slipping away

Running a studio comes with an invisible layer of work that nobody accounts for when they set their rate. Accounting, payment follow-ups, maintenance communications, project documentation, contracts. Real hours that don’t appear in any brief but consume the week.

That layer is now almost entirely automated. Not because it’s unimportant — it’s critical — but because it doesn’t require creative judgment. It requires precision and consistency, which is exactly what these tools do well.

The result is simple: more hours available for client work. Not to work more. To work better.

Production: custom tools for each project

In production, the most useful change wasn’t the most obvious one.

Repetitive tasks — asset processing, file renaming, exports, technical checks — can be automated and they are. But the bigger shift has been in code: I can now direct more complex builds with fewer resources, because I understand what I’m asking for and can bridge the gap between what a project needs and what a developer can deliver.

And I’ve been able to build project-specific tools that didn’t exist before. Not generic solutions — a version-tracking system for a long campaign, a production dashboard for an asset-heavy project, an internal validation flow before client delivery. Tools that previously would have cost more to build than they saved. Not anymore.

What changed about how I direct

This is the part I didn’t expect.

When the time that used to go into management comes back, you don’t just do the same things faster. You think differently. You show up to conversations with more context. You catch earlier when a project is drifting in the wrong direction. You can say no with better information.

Good direction isn’t just having good judgment. It’s having the mental space to apply it. That’s what I got back.

The limit

What I wouldn’t automate: the judgment itself.

What to propose, what to push back on, how to read a client, when a project needs to be redirected before it becomes a problem. The accumulated experience that makes a decision good. That has no shortcut, and it shouldn’t.

AI helps me arrive at that moment with more energy and better information. What happens in that moment is still mine.