README / README.md
thehonestape's picture
Repoint dataset link to brand-voice-spec
ed4fe35 verified
|
Raw
History Blame Contribute Delete
1.73 kB
metadata
title: Workhorse
emoji: 🐴
colorFrom: yellow
colorTo: gray
sdk: static
pinned: false

Workhorse

Strategic design partners for experts, entrepreneurs, and enterprise.

An independent design studio since 2014, Capitol Hill, Washington D.C. We design identity systems, build design infrastructure, and run a creative ops platform for AI-native brands.

Here on the Hub we publish the parts of that work that belong in the open: the layer between design systems and generative models.

The thesis

Brand is not decoration you apply at the end. It is the input. We keep tokens, voice, motion, and imagery direction in one machine-readable spec, and every output reads from it. Copy, video, social cards, generated images: same spec, different brand, visibly different result.

Most teams keep brand voice in a slide deck no model can load. We keep it as data, and we feed corrections back into it, so the brand compounds instead of rotting.

On the Hub

  • 📐 brand-voice-spec — the machine-readable format we use to keep AI copy on-brand, with a complete worked example. Registers, a lexicon, and a changelog that links every rule to the correction that produced it.
  • 🎬 ai-video-pipeline — text to short video. SDXL frames stitched with FFmpeg.

Work with us

We take on work where design systems meet generative models: brand-conditioned generation, style steering, preference data, and creative tooling for people who already know their craft.

Led by Abe Garcia. Studio: wrkhrs.co · abe@wrkhrs.co