Public site for Human Agent Interface (HAI) — a control layer where human intent stays inspectable before agentic execution.
Live: human-agent-interface.com
HAI is about making human decisions executable, inspectable, and recoverable when working with AI agents — not about pretending agents are autonomous.
human intent
→ context · routing · memory
→ scoped agent work
→ verification · handoff
→ owner decision
Static site / frontend for the public HAI presence (JavaScript). Browse the live site for the product narrative; this repo is the source behind it.
- Control-plane implementation (MCP):
smlfg/hai-mcp - Profile / build principles:
smlfg/smlfg
MIT — see LICENSE.