Abstract

The growth of open-source artificial intelligence tools and low-cost application development platforms has dramatically lowered the barrier for small and mid-sized public, nonprofit, and community-serving organizations to deploy custom software. Yet these organizations rarely pursue such solutions, because they lack either the vendor relationships or the internal technical capacity required for sustained AI implementation (Neumann et al. 2024). This TREO Talk introduces a replicable framework that positions university business schools—including non-IT departments—as implementation partners capable of coproducing AI-leveraged applications with community organizations through experiential learning. The framework has a three-phase development model designed to minimize risk to both parties while aligning with the academic calendar. Phase 1 focuses on Prototype development spanning 8–12 weeks at $0 cost to the partner. It is executed by supervised student capstone teams using open-source tools such as React Native, Firebase, and open-source LLMs. In Phase 2 we create Pilot minimum viable product (MVP) in 12–16 weeks, approximate cost is $15K–$25K here students harden the prototype into a deployable product through a dedicated team of advanced students, graduate assistants, and faculty. Phase 3 is Full Production, which takes 16–24 weeks and costs the partner ~$50K–$ 100 K. The phase scales the application with enterprise-grade infrastructure, accessibility compliance, and integration with the partner’s existing systems. The phased structure lets partners commit incrementally, only after observing functional evidence at each prior stage. Because the technology stack is entirely open source at the prototype stage, the framework is deliberately designed to be adoptable by departments without IT specialization—management, marketing, or accounting faculty, for instance, can lead initiatives by drawing on cross-disciplinary student teams. This broadens the pool of faculty champions and reframes AI implementation for societal impact as a scalable institutional activity rather than an isolated IS initiative. This work contributes to the AI implementation discourse along three dimensions. First, it identifies insights into the enablers and deterrents to AI adoption in resource-constrained organizations—around data governance, workforce adaptation, and stakeholder engagement across public and private sectors. Second, it offers a design science artifact: a process framework that is documentable, replicable, and comparable across institutional settings. Third, we generate findings on how experiential AI development shapes student learning potential, professional identity, and ethical reasoning. The research assesses three questions: RQ1: What factors best predict an effective community partner match for an AI project? RQ2: How should societal impact be measured beyond a single capstone term? RQ3: What governance tools for data management, intellectual property, and support protect partners once the university team delivers a product? Keywords: experiential learning; AI implementation; open-source AI; design science; university– community partnerships; IS pedagogy. References: Neumann, O., Guirguis, K., & Steiner, R. (2024). Exploring artificial intelligence adoption in public organizations: a comparative case study. Public Management Review, 26(1), 114–141. https://doi.org/10.1080/14719037.2022.2048685 experiential learning

Download

Share

COinS