SBIR Phase I: Integrating MentorAI into a student success platform

The broader/commercial impact of this SBIR Phase I project addresses the critical need for scalable, personalized student support in higher education. The project will develop an artificial intelligence (AI)-assisted mentoring platform that enhances peer mentoring programs through data-informed, evidence-based guidance. This innovation comes at a crucial time, as student distress rates have doubled over the past decade, and institutions struggle to meet growing demands for mental health and academic support. The technology will particularly benefit underrepresented students, who often face barriers accessing traditional support services. By combining AI capabilities with human peer mentors, this innovation will make technical advances in how to leverage AI tools within the context of human interactions. This will enable institutions to affordably scale high-quality, site-specific support services that improve student retention and success, advancing the health and wellbeing, academic achievement, and economic prosperity of marginalized students. The commercial potential is significant, with the mentoring software market projected to reach $1.3 billion by 2027. The platform's unique integration of data-driven insights with affordably scaled peer mentoring creates a competitive advantage in this growing market. The business model focuses initially on higher education institutions, with potential expansion into nonprofit, government, and professional development sectors. This product enhancement will offer unique features that address growing demands for personalized, evidence-based support. This Small Business Innovation Research (SBIR) Phase I project will develop and validate an innovative integration of large language models with retrieval-augmented generation technology to enhance peer mentoring effectiveness. The research addresses technical challenges in secure data integration, model fine-tuning, and scalable system architecture. The project will implement advanced encryption methods and differential privacy techniques to protect sensitive student information while enabling real-time, personalized support. The system architecture employs a modular, multi-tenant design that allows customization for specific institutional contexts while maintaining response times below 500 milliseconds. The research methodology includes developing secure protocols for data integration, implementing bias detection algorithms, and creating a comprehensive ethical framework for a "trustworthy knowledge-in-the-loop" approach using Retrieval-Augmented Generation technology to ensure accurate and evidence-based responses. Technical objectives include achieving 90% accuracy in contextually relevant responses and 85% user satisfaction ratings. The anticipated results include a fully operational prototype demonstrating secure integration of multiple data sources, personalized recommendation generation, and scalable performance under peak usage conditions This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.