I revisited some old 3D projects and was struck by how long it once took to build a scene - modeling, texturing, lighting, rendering. It suddenly felt almost retro, like shooting on film versus taking a digital photo.
That contrast sparked the premise: what happens when old-school methods - and mentalities - collide with a new way of creating? The mafia metaphor came naturally. Few archetypes resist change like old-school mobsters, and few tools embody acceleration like AI.
Writing comedy with AI quickly revealed something important: humor doesn’t live in the line itself - it lives in the delivery.
Just like in real filmmaking, you’re at the mercy of: tone, timing, reaction shots, anticipation, and surprise.
A punchline can be perfect on paper and completely fail depending on how it’s delivered - or what came right before it.
So instead of treating AI like a script generator, I treated it like actors in a casting session.For almost every joke or key line, I generated 10–12 different delivery options: different tones, different pacing, different emotional weight (face mimics, gestures).
Then, in the edit, I lego-ed those fragments together - assembling reactions, pauses, and beats until the scene felt right. This is exactly how comedy is built in real productions. The only difference is that here, the “actors” were simulated.
One of the biggest advantages of working with AI is something rarely talked about: you can build the film chronologically.
In traditional TV or film productions, scenes are shot out of order - based on locations, budgets, and logistics. You often don’t fully know how the story feels until everything is stitched together later.
With AI, I could:
That meant the story evolved like a real narrative - not a production puzzle. From the very beginning, I knew what the tone was, what should follow, and where the rhythm needed to change.
It felt closer to writing a short story than managing a shoot.