Stages
Houdini AI Assistant: Development Roadmap
▶️ Stage 1 – Initial Launch (v0.1.0)
Core features:
Primary AI integration (Gemini API)
Houdini scene analysis + contextual answers
Network and HDA generation from prompt (beta)
Built-in documentation + Houdini-style UI
Goal: attract first users + gather substantial feedback
▶️ Stage 2 – Multi-AI Integration (v0.1.5)
New features:
Support for GPT (OpenAI)
In-app AI selector
Modular system to quickly add other AIs (Claude, Mistral, ...)
Benefit: users can choose their preferred AI; broader appeal
▶️ Stage 3 – RAG (Retrieval-Augmented Generation) (v0.2.0)
New features:
Automatic import and indexing of all projects and HDAs from a folder
Advanced search + answers grounded in the user’s own docs and projects
Option to run RAG locally (no cloud)
Benefit: AI responds as if it were “trained” on the user’s projects
▶️ Stage 4 – Chain of Thought (CoT) & Mixture of Experts (MoE) (v0.2.5)
New features:
Advanced step-by-step reasoning mode (visible in the UI)
MoE → split tasks across specialized AIs:
AI for Houdini Python code
AI for VEX / wrangles
AI for procedural logic
AI for performance optimization
Benefit: more accurate and specialized answers
Last updated