About Grain
Grain is the universal interaction layer for AI interfaces. We believe that as AI systems become more capable and autonomous, we need a standard vocabulary for how they communicate with humans and with each other.
Our Mission
To create the standard markup language that enables seamless interaction between AI models, humans, and autonomous agents across every platform and interface.
The Problem We Solve
Every AI tool today rebuilds the same primitives:
- Chat UI
- Streaming text
- Tool calls
- Artifact rendering
- Human-in-the-loop approvals
This fragmentation wastes engineering time and creates siloed ecosystems.
Our Solution
Grain provides a unified, declarative language that:
- AI models can output
- Any platform can render
- Works across Web, CLI, MCP, and Agent-to-Agent
Just as HTML standardized document rendering, Grain standardizes AI interaction.
What "Spec-First" Means Here
Grain is built around a simple rule: public examples are not decorative. If a parser, renderer, or adapter disagrees with the docs, the implementation is wrong or the docs must be updated in the same change.
That is why the docs site lives alongside the packages and ships with its own build and preview workflow. The language reference, package READMEs, and homepage playground are part of the release surface.
The Meridian Philosophy
Grain operates on the "Meridian Layer"—the boundary between machine intelligence and human experience.
By formalizing this layer:
- AI models focus on reasoning
- Platforms focus on rendering
- Users get consistent experiences
Open Source
Grain is fully open source under the MIT license.
We believe in building transparent, community-driven tools that anyone can use, modify, and contribute to.
Team
Grain is built by a team of engineers and designers passionate about the future of human-AI interaction.
We come from backgrounds in:
- AI/ML infrastructure
- Developer tools
- Open source software
- UX design
Current Focus
The current focus is practical production readiness:
- Keep the Grain language contract aligned with the parser and validators.
- Ship stable adapters for web, CLI, and MCP surfaces.
- Make the documentation auditable, previewable, and deployable as a first-class product surface.
Contact
- GitHub: github.com/sir-ad/grain
- Issues: github.com/sir-ad/grain/issues
- NPM: @grain.sh/core
Acknowledgments
Grain draws inspiration from:
- HTML/XML for the markup approach
- React for component thinking
- XState for state machine patterns
- The broader AI/ML open source community
© 2026 Grain Language. Built for the era of intelligence.