Docs/CHINNA_AI.md
← All docs
MarkdownHTML Render

Chinna-AI

The intelligence behind every Fantom surface — see, imagine, build, live.

Chinna-AI

The intelligence behind every Fantom Interio surface — one model context shared across interior design, planning and the live home.

What Chinna-AI does

Layer Capability
See Room analysis, floor-plan parsing, style and material recognition
Imagine Layout-safe photoreal redesign (grok-imagine batches: 6 standard / 4 pro), reference transfer, chat editing
Build 3D scene graphs, device placement, wiring routes, INR quotations
Live Tool-calling device control, scenes, energy routing, security posture

Where it runs

  • Interior tools/dream, /editor, /studio: prompts composed by lib/ai/promptEngine.ts with STRICT_RULES (architecture frozen, perspective frozen, minimal-change interpretation).
  • Smart Home OS/smart-home-hub: the assistant calls real tools (toggle_device, run_scene, set_all_lights, show_widget, …) through /api/hub/assistant and renders generative UI widgets inline.
  • Offline-first — a deterministic local intent engine handles common commands with no cloud at all; Alexa and Google remain optional bridges, never the core.

Principles

  1. Human approval before irreversible action.
  2. The project history stays visible — every generation, cost and revision.
  3. Outputs are proposals, not certification.