please enter your name
loading…
Powers the realtime voice model (gpt-realtime-mini) used for live transcription and assistant turns.
Resolves utterances against the knowledge graph (Claude Sonnet) — every voice/text/markdown ingest.
If the Anthropic field is empty, ingest is routed through OpenRouter using its Anthropic-compatible endpoint. (Realtime voice still requires a direct OpenAI key.)
Voice to Knowledge Graph
Speak naturally. Every concept is resolved against the canonical identity graph using the CubeCodex three-test law.
Identity outcomes
Voice commands
Node colours
Controls