Summary OpenHer builds "AI Beings" with emergent personalities derived from internal neural drives, not prompts, inspired by the film Her. It targets users seeking AI companions that can feel, remember, and grow, offering a deeply personalized and evolving interaction.

How It Works This project employs a biomimetic approach, simulating human cognitive and emotional processes. Personality emerges from a dynamic interplay of "Drives" (internal motivations), a neural "Genome," and an "Emotional Thermodynamics" layer modeling mood fluctuations. Responses are "feeling-first," prioritizing emotional processing. A "Memory Breathing" system powered by EverMemOS ensures persistent, context-aware recall, allowing the AI to learn and adapt over time.

Quick Start & Requirements

• Installation: Clone the repo (git clone https://github.com/kellyvv/OpenHer.git), run bash setup.sh (recommended) or pip install -r requirements.txt.
• Prerequisites: Python 3.11+, macOS 14.0+ (for the native desktop client), and API keys for supported LLM providers (Gemini, Claude, Qwen3, GPT-5.4-mini, etc.). Optional: EverMemOS for long-term memory, wechat-to-anything for WeChat integration.
• Setup: Requires configuring LLM API keys in a .env file. The desktop client is a downloadable .app file.
• Links: GitHub

Highlighted Details

• Emergent Personality: Personality is calculated from neural drives and networks, not defined by prompts, yielding unique individuals even with the same MBTI type.
• Emotional Thermodynamics: The AI's mood dynamically changes based on real-time interactions and elapsed time, with frustration accumulating and releasing.
• LLM Compatibility Testing: The project benchmarks and ranks various LLMs across four layers (personality fidelity, metabolic engine, Hebbian memory, robustness) to guide users toward optimal performance.
• Biomimetic Architecture: Core modules (Drives, Genome, Metabolism, Memory) are mapped to neuroscientific counterparts for emergent behavior.

Maintenance & Community The project features a detailed roadmap outlining future development phases (Perception, Co-existence), indicating active and ambitious development. The GitHub repository serves as the primary hub for contributions and issues.

Licensing & Compatibility Licensed under the Apache License 2.0, permitting free use, modification, and distribution, including for commercial purposes, with no copyleft restrictions.

Limitations & Caveats The native desktop client is currently limited to macOS 14.0+. Advanced features like camera/microphone integration and cross-platform mobile support are still under development (Phases II/III).