| # Responsible AI Initiative |
| |
| The Responsible AI Initiative (RAI) advances the ASF's mission of building, |
| maintaining, and releasing software for the public good in the age of generative |
| AI. RAI supports Apache projects and the broader ASF community. |
| |
| ## What RAI provides |
| |
| - **Access to AI models and tooling** — giving the Foundation and its projects |
| access to language and code models for experimentation and integration across |
| core services including Security, Tooling, and Infrastructure. |
| - **Project-level ecosystem support** — helping ASF projects accelerate |
| production-ready AI development with proven, widely adopted technologies across |
| the full AI/ML stack. |
| - **Community engagement** — learning, collaboration, and contribution through a |
| Responsible AI track at Community Over Code, hackathons, meetups, and industry |
| events. |
| |
| ## What we're building |
| |
| The systems are taking shape as project committees: |
| |
| - an **LLM gateway and model management** layer, |
| - **agent harnesses**, and |
| - **MCP server frameworks**. |
| |
| These are the pieces of the platform the [roadmap](roadmap/) lays out, organized |
| around a single architecture. |
| |
| ## Where things live |
| |
| - [roadmap/](roadmap/) — the proposed platform architecture. |
| - [../platform/](../platform/) — setup and runbooks for the systems the |
| architecture depends on (agent runtime, self-hosted model serving). |
| - [../security-pipeline/](../security-pipeline/) — the security audit pipeline. |
