FLOWOOD, Miss. — Dr. Kollin Napier, chief artificial intelligence officer for the Mississippi Artificial Intelligence Network (MAIN), presented a practical model for helping state agencies test artificial intelligence ideas without treating a prototype as a finished system. His September 9 session at the 2026 Mississippi Digital Government Summit focused on how Mississippi’s AI Innovation Hub turns a bounded public-sector problem into a governed proof of concept and evidence for an agency’s next decision.
The session, “Bits and Bytes — Mississippi’s AI Innovation Hub: From Ideas to Impact,” was held at the Sheraton Flowood The Refuge Hotel & Conference Center. Napier described the Hub as a repeatable way for government, higher education and industry to learn together while keeping human oversight and agency responsibility at the center of the work.
A governed path from an agency problem to a decision
The Mississippi AI Innovation Hub begins with a defined operational problem, not a request to deploy AI broadly. An agency identifies the intended users, the data or systems involved, the result it needs and an accountable owner. The proposed use case then moves through a governance and readiness review before a student team begins work.
The Hub’s operating model has five stages:
- Intake: Bound the problem and identify the agency owner.
- Review: Assess risk, data readiness and whether a proof of concept is appropriate.
- Team formation: Match students and mentors to the approved challenge.
- Sandbox build: Develop and test the prototype in a governed environment.
- Handoff: Demonstrate the work, document what was learned and help the agency decide what should happen next.
Typical proof-of-concept work lasts six to eight weeks. The output is evidence about feasibility, data needs, safeguards and operational fit. A prototype is not an authorization to deploy, and any production decision remains with the agency.
Eight project briefs across seven state agencies
Presentation materials current through September 3 showed eight 2026 project briefs spanning seven state agencies and three university partners: Mississippi State University, the University of Southern Mississippi and the University of Mississippi. The Hub had also received more than 20 submitted use cases. Napier distinguished that larger number as a pipeline of possible projects rather than a count of completed builds.
The portfolio covers areas including procurement support, certificate and compliance workflows, document analysis, nutrition-service access, transcript review and internal help-desk assistance. The common thread is a narrow use case with a defined public-service purpose and a process for evaluating the result.
The partnership aligns three distinct roles. The Mississippi Department of Information Technology Services (ITS) provides governance, the cloud landing zone and agency intake. Amazon Web Services (AWS) supports proof-of-concept funding, landing-zone credits and learning resources. MAIN provides education, workforce preparation and statewide coordination.
Prototypes show both promise and limits
Napier used several projects to show what an evidence-producing prototype can look like.
Project C.A.I.R.O. explored certificate lifecycle work involving ITS and the Mississippi Development Authority. According to the project’s reported demonstration, the prototype identified 15 of 15 certificates that needed renewal and analyzed a set of 20 certificates in about 20 seconds. The public repository also makes the limits clear: the work is a proof of concept, some steps are simulated, and production use would require additional security, privacy, testing and stakeholder review.
A separate transcript-verification concept examined how AI could extract information, run checks and flag records for review. Napier emphasized that a high-stakes workflow still requires people to examine the evidence and make the decision.
Procurii, created by Mississippi State University students for internal ITS use, explored an AI assistant for procurement guidance. The proof of concept was designed to draw from official ITS procurement materials, cite those sources and support document-version tracking and retraining. It remains an advisory prototype, not a substitute for official procurement decisions.
Public code and documentation for these and other projects are available through the MS ITS AI Innovation Hub organization on GitHub. The repositories give agencies, educators and students a way to examine the work while retaining the safeguards and limitations documented for each project.
One project, two forms of public value
The model is designed to produce value for agencies and students at the same time. Agencies receive a tested concept, documented findings and clearer evidence for deciding whether an idea merits further investment. Students work with real stakeholders, team constraints, governance requirements and public-sector problems while creating project artifacts they can discuss with future employers.
That dual return helps connect responsible technology experimentation with Mississippi’s workforce needs. It also gives agencies a lower-risk way to learn what a proposed tool can and cannot do before considering a larger implementation.
Governance begins before the build
Napier connected the Hub’s intake process to Mississippi’s broader work on AI readiness. Before a team begins building, the agency and project reviewers must be able to explain the problem, the intended users, the data involved, the accountable owner and the decision the prototype should inform.
This approach complements MAIN’s voluntary statewide AI competency framework, which helps organizations consider the knowledge and responsibilities people need when they use or oversee AI. The framework does not replace agency policy or professional standards.
Bring a bounded challenge to the Hub
Mississippi state agencies interested in exploring a suitable use case can review the AI Innovation Hub process and project portfolio. A strong submission identifies a specific public-service problem, an agency owner and a result that can be examined within a short proof-of-concept cycle.
Learners and organizations can also explore MAIN’s current catalog of no-cost, online and self-paced AI courses, which support foundational skills and more specialized learning. Together, the Hub and the learning catalog provide practical entry points for agencies and individuals preparing to use AI responsibly.