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

arXiv cs.AI / 4/20/2026

💬 OpinionSignals & Early TrendsTools & Practical UsageModels & Research

Key Points

  • The paper introduces SocialWise, a browser-based LLM-agent conversational therapy system aimed at improving communication skills for people with Autism Spectrum Disorder (ASD).
  • SocialWise combines LLM conversational agents with a therapeutic RAG (retrieval augmented generation) knowledge base to provide scenario-based practice (e.g., ordering food or joining a group).
  • Users can interact via text or voice and receive immediate, structured feedback focused on tone, engagement, and suggested alternative phrasings.
  • The prototype is implemented using Streamlit, LangChain, and ChromaDB, and is designed to run on any internet-connected computer to offer scalable, on-demand coaching.
  • The work positions recent LLM advances as a way to address the limited availability and high cost of traditional, in-person role-play therapy while improving everyday conversation practice.

Abstract

Autism Spectrum Disorder (ASD) affects more than 75 million people worldwide. However, scalable support for practicing everyday conversation is scarce: Low-cost activities such as story reading yield limited improvement. At the same time, effective role-play therapy demands expensive, in-person sessions with specialists. SocialWise bridges this gap through a browser-based application that pairs LLM conversational agents with a therapeutic retrieval augmented generation (RAG) knowledge base. Users select a scenario (e.g., ordering food, joining a group), interact by text or voice, and receive instant, structured feedback on tone, engagement, and alternative phrasing. The SocialWise prototype, implemented with Streamlit, LangChain, and ChromaDB, runs on any computer with internet access, and demonstrates how recent advances in LLM can provide evidence-based, on-demand communication coaching for individuals with ASD.