SocialWise: LLM-Agentic Conversation Therapy for Individuals with Autism Spectrum Disorder to Enhance Communication Skills

Researcher Albert Tang has introduced SocialWise, a novel application that leverages Large Language Model (LLM) conversational agents to provide scalable communication therapy for individuals with Autism Spectrum Disorder (ASD). The tool addresses a critical gap for the over 75 million people affected globally, where effective role-play therapy is often limited by the high cost and scarcity of specialists. SocialWise operates as a browser-based app, making it accessible on any computer with internet access.

The system is built using a modern AI stack including Streamlit for the interface, LangChain for agent orchestration, and ChromaDB for its vector database. At its core is a therapeutic retrieval-augmented generation (RAG) knowledge base, which grounds the AI's responses in evidence-based practices. Users can select from various everyday social scenarios—such as ordering food or joining a group—and interact with the AI agent through text or voice input.

During these simulated conversations, SocialWise provides instant, structured feedback on key communication elements like tone, engagement level, and suggests alternative phrasing. This on-demand coaching model demonstrates how recent advances in LLM capabilities can be harnessed to deliver personalized, therapeutic support outside of traditional clinical settings, potentially democratizing access to essential social skills training.