The language of your AI layer

🚀 from idea to demo to prod Write low-code config in AI JSON and AI YAML.

alright show me what it looks like

Hyper Readable

no, really, if you have ideas on how to make it more readable, let's talk

Features

Built with ❤️ for teams who move fast

Rapidly iterate ideas, demonstrate value, and productionize at scale.

Platform Agnostic

Use any language model, any config format, and deploy anywhere. We support files with the .ai.json and .ai.yaml file extensions, with .ai.toml and .ai.pkl on the roadmap.

VSCode Language Server

Autocomplete, syntax highlighting, and native jinja support. Catch and fix semantic errors before ever pressing play.

Real-Time Demo

Launch a demo of your pipeline straight in your IDE. As you iterate flow, the demo reloads automatically.

Meticulously Monitored

Run in production with the peace of mind that your application is structurally logged, with the ability to trace any error through its execution.

Open Source (MIT)

AI JSON is an open standard for expressing AI pipelines. We will only ever make our licensing less restrictive.

Semantically Versioned

Each pipeline specifies the version of AI JSON it was written for; we aim to provide automatic migrations between minor versions of the library.

What AI flows are you building?

Get in touch, we'd love to help.