AI-powered Chatbot: The National Council for Mental Wellbeing

The Arizona State University Artificial Intelligence Cloud Innovation Center, powered by Amazon Web Services (AWS), collaborated with The National Council for Mental Wellbeing to design an AI-powered chatbot to be integrated into the Mental Health First Aid (MHFA) Learning Ecosystem. The solution provides natural language access to a centralized knowledge base of MHFA training materials, enabling instructors and internal staff to retrieve accurate, citation-backed guidance in real time. By combining Retrieval-Augmented Generation (RAG), bilingual support in English and Spanish, role-based personalization, and transparent source citations. The AI-powered chatbot improves access to trusted information while reducing administrative burden and operational complexity.

The National Council for Mental Wellbeing supports more than 3,200 organizations and has trained over 4.7 million individuals in its Mental Health First Aid program. As demand for MHFA training continues to grow nationwide, instructors and staff face increasing difficulty accessing consistent, up-to-date training materials across fragmented systems and static documents. Manual support processes, reliance on PDFs, and ad hoc communication channels create delays, inefficiencies, and uncertainty. Additionally, the organization has limited visibility into frequently asked questions, emerging trends, and knowledge gaps that could inform program improvements. Without a scalable, intelligent support system, the National Council risks slower response times, inconsistent guidance, and reduced capacity to meet rising mental health education needs.

The CIC team worked with the National Council to design the AI-powered chatbot as a secure, scalable conversational support tool. The solution includes:

• AI-powered chatbot integrated into the MHFA Learning Ecosystem
• Centralized knowledge base built from approved MHFA materials
• Retrieval-Augmented Generation (RAG) using Amazon Bedrock with citation-backed responses
• Vector-based semantic search via OpenSearch Serverless
• English and Spanish bilingual support with seamless switching
• Role-based personalization behind secure login
• Real-time response streaming for interactive experience
• AI-driven escalation to Zendesk for low-confidence queries
• Admin dashboard for document management, logs, and analytics
• Secure authentication and conversation storage using Cognito and DynamoDB

The AI-powered chatbot transforms fragmented training documentation into a conversational, searchable, and citation-backed knowledge experience. The solution supports MHFA instructors and staff by providing quick access to official guidance, reducing repetitive administrative inquiries, increasing confidence in training delivery, and enabling data-driven insights into knowledge gaps and frequently asked questions. By improving efficiency, transparency, and scalability, the AI-powered chatbot strengthens the National Council’s ability to expand Mental Health First Aid training nationwide while maintaining quality and consistency.

“Our collaboration with AWS CIC has accelerated the modernization of our learning platform by integrating AI to better serve instructors, learners, and support teams. The student team delivered a thoughtful, high-quality solution that improves access to trusted

 information today and lays out the groundwork for future modernization and scalable learning. These advancements support our mission to expand access to high-quality, equitable mental health and substance use education for all.” 

Sunita Uthra | Vice President, Enterprise Technology | National Council for Mental Wellbeing

While developed for the MHFA Learning Ecosystem, the AI-powered chatbot architecture can extend to additional membership organizations, certification programs, and professional training ecosystems that require secure, citation-backed knowledge retrieval. The modular serverless design enables future CRM integrations, advanced analytics dashboards, multilingual expansion, and deeper personalization features to support large-scale educational delivery models.