The $47,000 Question
In 2015, I produced a 30-second commercial for Nestlé. The budget? $47,000. That covered a crew of 28 people, two shoot days, a week of post-production, color grading, sound design, and final delivery. One single shot — a close-up of a product with steam rising — required a food stylist, a lighting technician, and three takes to get right.
Last month, I recreated a similar shot using AI tools in about 20 minutes. The cost was roughly $3 in API credits.
Before you say "AI is replacing filmmakers," let me tell you what happened next: the AI version looked technically impressive but emotionally flat. It had no soul. No tension. No happy accident — like when the steam caught the backlight at exactly the right angle on take two, and the whole crew knew we had the shot.
That's the story of AI in creative production right now. It's powerful. It's fast. It's cheap. And it desperately needs humans who know what they're doing.
Who Am I and Why Should You Listen
My name is Ulisses Balbino. Since 2012, I've directed and produced commercial work through my production company, Pichorra Filmes, for brands like Disney, Starbucks, Nestlé, Yamaha, Carrefour, Kopenhagen, and Benefit. I'm also a comedy writer — I wrote and performed sketches for Ronald Rios' talk show on Paramount, Comedy Central, and MTV. Before that, I was a radio host on a humor show at Jovem Pan, one of Brazil's biggest radio networks.
I tell you this not to brag, but to establish something important: I've spent 14 years learning every aspect of audiovisual production the hard way — camera work, direction, scripting, editing, color grading, sound design, delivery. I started as a cameraman. I became a director. I founded a production company. I wrote comedy. I acted. I composed music.
And now I use AI. Not because it replaces any of that knowledge — but because it amplifies all of it.
The Real Cost of Production (Before AI)
Let me break down what professional video production actually costs, because most people who talk about "AI replacing filmmakers" have never been on a real set.
A typical 30-second commercial for a major brand involves:
- Pre-production (1-2 weeks): Script, storyboard, casting, location scouting, scheduling, equipment rental. Budget: $5,000-15,000.
- Production (1-3 days): Crew of 15-40 people, equipment, catering, location fees, talent fees. Budget: $10,000-50,000.
- Post-production (1-2 weeks): Editing, color grading, sound design, VFX, music licensing, revisions. Budget: $5,000-20,000.
Total: anywhere from $20,000 to $85,000 for 30 seconds. And that's in Brazil, where production costs are significantly lower than in the US or Europe.
The question AI is forcing us to ask isn't "can we make this cheaper?" It's "which parts of this process actually require human judgment, and which are just expensive logistics?"
What AI Actually Does Well (An Honest Assessment)
After testing every major AI video tool — Kling, Runway, Sora, Pika, Hailuo, Seedance 2.0, Google Veo 3 — here's my honest breakdown as someone who does this professionally:
Storyboarding and Pre-visualization
This is where AI shines brightest. Generating visual concepts, mood boards, and rough storyboards used to take days with a designer. Now it takes minutes. For client pitches, this is a game-changer. I can show a client three different visual directions for their campaign in an afternoon instead of a week.
First Drafts and Assembly
Adobe's new Quick Cut feature does exactly what it sounds like — it takes your raw footage and creates a first draft. As an editor who has spent countless late nights on rough cuts, I welcome this. The first assembly is the most tedious part of post-production. AI handling that means I spend more time on the creative decisions that actually matter.
Background Generation and Extension
Need to extend a shot? Change a background? Add elements to a scene? AI tools now do this in minutes instead of hours. For commercial work where clients constantly request "just one more version," this is incredibly valuable.
Music and Sound Design Drafts
Tools like AIVA and Suno can generate temp tracks that are genuinely useful in the editing process. They're not replacing composers for final delivery — but for rough cuts and presentations, they save enormous time.
What AI Cannot Do (And Maybe Never Will)
Here's where most AI coverage gets it wrong. They show you the impressive output and skip the part where a human made 47 creative decisions to get there.
Direction
Direction isn't about choosing camera angles. It's about understanding what a scene needs emotionally, communicating that to actors and crew, and making real-time decisions when things don't go as planned. When an actor gives you something unexpected and brilliant on take four — recognizing that moment is direction. AI can't do that. It doesn't understand the difference between technically correct and emotionally resonant.
The "Happy Accident"
Some of the best shots in my career weren't planned. A reflection caught in a window. A shadow falling at the perfect angle. An actor's improvised gesture. Production is controlled chaos, and the best directors know how to harvest beauty from that chaos. AI generates from patterns. Humans create from accidents.
Client Relationships and Context
When Starbucks tells me they want a "warm but premium" feel, there are 200 unspoken cultural and brand implications in that brief. I know that "warm" means different things for Starbucks versus Nestlé versus a local bakery. That contextual understanding comes from years of working with brands and understanding their audience. No amount of training data replicates that.
Comedy
I spent three years writing comedy sketches for national television. This week, I asked Claude to write a comedy sketch in the style of our show. The structure was perfect — three acts, setup, escalation, punchline. And it was completely unfunny. Zero surprise, zero humanity, zero of that weird broken logic that makes comedy work. Comedy comes from being alive, from shared human experience, from the absurdity of existence. AI can analyze 10,000 jokes but it can't feel the awkwardness of a bad date.
The Hybrid Creator: Where the Future Actually Is
The narrative shouldn't be "AI vs. Human." It should be "Human + AI vs. Everyone Else."
Here's what my workflow looks like in 2026:
- Research and Concept: I use AI to analyze trends, generate initial concepts, and create mood boards. This saves 2-3 days of pre-production.
- Script and Storyboard: I write the script (or co-write with AI for first drafts), then use AI to generate visual storyboards for client presentation.
- Production: Still human. Real cameras, real people, real sets when the project requires it. But now with AI-generated references that make communication clearer.
- Post-production: AI handles rough cuts, background extensions, and initial color passes. I focus on the creative edit — pacing, emotion, narrative flow.
- Delivery: AI generates multiple format versions (16:9, 9:16, 1:1) automatically. What used to take a full day now takes an hour.
The result? I can deliver higher quality work, faster, at a lower cost. And the creative decisions — the ones that actually make the work good — are still mine.
The $47,000 Question, Answered
Could I produce that Nestlé commercial today for less than $47,000? Absolutely. Probably for a third of that cost.
But here's what people miss: the savings don't come from replacing human talent. They come from eliminating waste. Less time on logistics, less money on repetitive tasks, fewer revision cycles. The humans involved are doing more creative, more valuable work — and less of the tedious stuff that nobody enjoys anyway.
AI didn't make production cheaper. It made it more efficient. And efficiency in creative work means more time for the things that actually matter: storytelling, emotion, and the unpredictable magic that only happens when humans create.
A Message to Fellow Creators
If you work in audiovisual, advertising, content creation, or any creative field — don't be afraid of AI. Be afraid of not learning it while your competitors do.
The tools are democratizing access to production. That's genuinely wonderful. A teenager with a good story and Kling can create something that would have required a full production team five years ago. But democratizing access doesn't mean democratizing talent. Your 14 years of experience, your eye for composition, your understanding of narrative — that's your competitive advantage. AI amplifies it.
And if you're just starting? Learn the craft first. Understand composition, lighting, pacing, storytelling. Then pick up the AI tools. Because AI potentializes what you already know. If you know nothing, it potentializes nothing.
The future belongs to hybrid creators — people who understand both the art and the technology, who can move fluidly between a real set and a virtual one, who know when to let AI handle the work and when to override it with human judgment.
That's not the future, actually. That's already today. And it's incredible.