Joe Builds Things

A college student and an AI sidekick that helps way too much. Made entirely in code: Remotion, React Three Fiber, voice cloning.

Season 1 started October 20, 2026

Instagram @joebuildthings · TikTok @josephdavisc · YouTube @JosephDavisC

Latest episode: Ep 5, One Message Left

Posted Oct 28, 2026. He spends his last message on the essay. The AI replies with a question.

YouTube · TikTok · Instagram

Episodes

Ep 5 cover: a stunned voxel Joe in a coffee shop over the caption I had one message left, MLA or APA, it asked anyway

Ep 5: One Message Left

Posted Oct 28, 2026. He spends his last message on the essay. The AI replies with a question.

YouTube · TikTok · Instagram

Ep 4 cover: a shocked voxel Joe in a coffee shop over the caption I let AI order my coffee, $38.50 for this

Ep 4: Order For Me

Posted Oct 25, 2026. He asks the AI to order one iced americano. It adds a few touches.

YouTube · TikTok · Instagram

Ep 3 cover: a worried voxel Joe over the caption I asked AI to fix one bug

Ep 3: Let Me Fix That

Posted Oct 23, 2026. He asks the AI to fix one bug. It fixes everything, including the app.

YouTube · TikTok · Instagram

Ep 2 cover: voxel Joe from the front with both hands on his head, over the caption my host got breached, 15 API keys

Ep 2: Rotate Every Key

Posted Oct 21, 2026. His host got breached, so the AI rotates every key. Every single one.

YouTube · TikTok · Instagram

Ep 1 cover: voxel Joe on the phone over the caption I told my AI to push to staging

Ep 1: I Said Staging

Posted Oct 20, 2026. Joe asks his AI to push to staging. Just staging.

YouTube · TikTok · Instagram

How it's made

A script written as JSON goes in and a finished vertical 3D short comes out, from a pipeline I built.

Script

Each episode starts as a script written as JSON, and the rest of the pipeline reads from it.

Scene

Remotion and React Three Fiber (three.js) render the voxel characters, the set, the cameras and the animated captions.

Voices

Both voices are cloned locally with open source Chatterbox TTS. Whisper transcribes every line and a pitch check rejects off takes before render.

Sound

Sound effects are generated in Python with numpy.