This sandbox environment lets developers build AI assistants that display human-like emotional responses during interactions. The system uses what Soul Machines calls Experiential AI, a patented approach that combines cognitive modeling with embodied cognition to generate reactions that mimic human empathy and emotional understanding.
The technical architecture is LLM agnostic. Developers can train agents on proprietary language models or integrate with existing models like GPT. This system doesn't lock you into a specific AI backend, which matters for teams with existing model investments or specific performance requirements.
The agents operate autonomously rather than following predetermined conversation scripts. This means each customer interaction follows a unique path based on real-time conversational context. The system processes input, generates emotional responses through its cognitive modeling layer, and produces output that packs both verbal responses and emotional expressions tailored to the conversation flow.
Customization covers visual appearance, conversational tone, and personality traits. Developers configure these parameters to align with brand requirements and specific business objectives. This system handles the underlying emotion generation while exposing controls for behavioral tuning.
The data pipeline integrates with GPT and other leading language models, though specific model names beyond GPT aren't documented. The system processes conversational input through the chosen LLM, then applies its proprietary emotional layer before generating the final response.
Technical limitations are significant. Soul Machines Limited entered receivership on February 5, 2026, with receivers appointed from KPMG New Zealand Limited. The company isn't providing normal services until further notice. This receivership status creates uncertainty about system availability, ongoing support, and long-term viability for any implementations built on this technology.
