AI research studio · New York

Make the implicit explicit.

Agents do their best work in discovery, when the explicit rules and instructions they need are documented where they can find them. Baseline publishes open-source tools and frameworks to support this.

Studio

A research thread, two ways.

The thesis: AI tools work best when the things they need to know are made explicit, not left implicit.

Baseline Core makes product methodology and business context explicit, so AI tools read, follow, and respect both, instead of inventing their own.

Runtime Classifier makes creative style explicit, so models don't need fine-tuning to follow a documented aesthetic.

Both projects ship open source under MIT. Free to use, free to fork.

Open-source projects

What's shipped.

Baseline Core

Your business, wired into your AI tools.

A product system for any AI tool.

12 skills, 14 frameworks, and 34 reference files covering research, strategy, specs, design, and planning. Delivered via CLI, loaded into any AI tool that reads AGENTS.md.

Open sourceMITNode
Runtime Classifier

Encode style. Skip the fine-tune.

A framework for explicit creative style.

An 8-dimension framework that encodes creative style as runtime rules instead of fine-tuned weights. Worked photography example. Translates to writing, music, art, and graphic design.

Open sourceMITPython

More projects in development.

Built by

Trent Mitchell

Designer, builder, and the person behind Baseline Studio. Based in New York. Currently building open-source tools and projects at the seam where creativity and AI meet.

If a project here is useful, or you have ideas for the next one, reach out.

Contact

Get in touch.

Feedback on a project, ideas for the next one, or just saying hello. I read everything.

trent@baselinestudio.design