Rnraside Audio Theatre
A human-led, AI-assisted audio drama, written, performed, and produced by Rodrigo Pool.
Listen the first time for the jokes, come back for the plot.
Watch: Fogbound, animated
The first 15 minutes of Fogbound, animated. Every facial performance is my own face capture, driving AI-generated faces in scenes I built in Blender. Released November 13, 2025.
Listen
How it’s made
Excuse Us is a one-person production. Every story, character, and creative decision is mine. AI tools handle specific support jobs, the way a small crew would on a bigger show.
| Stage | Rodrigo | AI tools |
|---|---|---|
| World and lore | Builds the world of Blind Corp | Claude tracks lore, references, and the show bible |
| Writing | Stories, scripts, and dialogue | No AI |
| Script formatting | Not involved | Claude |
| Character performance | Performs the voices | No AI |
| Narration and intros | Writes all of it and voices most of it | ElevenLabs, only where a passage needs it |
| Editing, sound, and mix | All of it | No AI |
| Animation | Animates in Blender, Animate, After Effects, and Premiere | Generative animation in select moments |
| Animated episodes (Fogbound, pilot) | Face-capture performance; bodies, sets, and scenes built in Blender | AI-generated faces, animated from my capture with LivePortrait, an open-source model I run locally |
| Artwork and website | Designs and directs | ChatGPT, Gemini, and Adobe generative AI |
Stages run in production order. An empty lane means that stage has no AI involvement, or no human involvement.
When the narrator is AI
I voice most of the show myself, narration included. Some passages need a cleaner delivery than my accent and diction allow. For those, I write the lines and direct an AI voice to perform them. The characters always stay human.
Fogbound: my performance and AI faces
Fogbound is a two-part story in Season 2. I animated its first 15 minutes. The characters’ faces are generated with AI and animated from my own face-capture performance. Everything else, from the bodies to the sets and scenes, I built in Blender. I used the same approach for a clip from the pilot episode.
The faces are synthetic, but every glance, beat, and reaction is mine.
The facial animation runs on LivePortrait, an open-source model I run on my own machine. Since making Fogbound, I’ve modified it, replacing its face detector with Google MediaPipe so my animation pipeline runs only on openly licensed components.
What AI doesn’t do
AI doesn’t write my stories, invent my characters, or give my performances. It keeps my world consistent, formats my pages, voices select narration, and adds generated visuals and animation where they lift the work. The creative judgment stays with me.
Writing a show with this many characters, timelines, and running jokes meant my notes were scattered everywhere. So I designed my own scriptwriting software, Script Room, around the way I actually write, and built it with Claude.
At its center is an interactive show bible. Every character gets a profile page I fill in as I go: their history, what they want, how they sound, and how they look. Scene breakdowns map out what happens in each episode before I write a line of dialogue. Lore gets tracked as it’s created, so a detail mentioned in Season 1 is still right when it matters in Season 3. All of it feeds an encyclopedia of the characters and world of Excuse Us, from Blind Corp to the last diner in Joiceville.
When it’s time to write, Script Room handles the formatting, so I can open a page and start typing the scene instead of fighting the layout.
I decide what the tool needs to do and how it should work. Claude writes and updates the code. When my workflow changes, I describe what I need, and the tool changes with it. It’s software shaped entirely around one writer’s process, and I’m developing it into something other writers can use.
Behind the scenes
Blender: Staging a Fogbound scene. Characters, backdrop, cameras, and lights are arranged in 3D space, so I can block each shot like a set, choosing angles and moving the camera through the room.
After Effects: The Fogbound title sequence. Typography, textures, a cracked-glass overlay, and a 3D camera move, designed and animated frame by frame.
Premiere Pro: Editing an animated sequence. More than a dozen video layers combine my animated character shots, generated stills, b-roll, and effects, all cut to the audio of the episode.
Audition: The audio session for a Season 2 episode. Over twenty tracks of dialogue, music, radio, and sound effects, each placed and mixed by hand. This is where the show actually gets built.