Unlocking Efficiency: Leveraging AI in Public School District Operations

Unlocking Efficiency: Leveraging AI in Public School District Operations

Ashley Murphy and Kris Hagel give this session at the ASBO International Annual Conference & Expo (AC&E 2026) on October 15, 2026, for school business officials. It covers privacy-first AI governance, practical prompting habits, the platform and agents Peninsula School District runs on its own AWS cloud, and a set of back-office examples: transportation routing, financial reconciliation, enrollment reporting, staff evaluations and budget dashboards.

Key Takeaways:

• Privacy sets what is safe to do: The session grounds AI use in PSD's data security guidance, which sorts tools into Sandbox, Lifeguard, Vault and Pocket AI by how much protection they give student and district data.
• A district-run platform: AI Studio is PSD's custom, open-source AI platform in its private AWS cloud. It lets the district build tools that match its instructional and operational goals, and gives AI agents a safe place to create content.
• Agents for staff: More than 250 staff have a personal PSD AI Agent running on AWS, with 30+ skills and integrations: role-based access to the PSD data warehouse, Google Workspace, Atrium for creating artifacts, and scheduled or proactive work. Staff talk to it through Google Chat, and what it can reach depends on their role.
• Routing a disputed route: A district "Pony" driver, a union position, said the route could not be finished in one eight-hour shift. Asked for the most efficient route to the district's 17 buildings with a five-minute stop at each, the AI said it could not calculate precise routes and recommended tools such as RouteXL instead, then formatted the addresses for import. The result showed the route could be done in four hours, that the existing route was not optimized, and gave an optimized route with stop times that could be supported with the union.
• Financial reconciliation: Staff gave the agent the county treasurer's monthly report and the WSIPC monthly trial balance to compare, then asked it to read the F-197 balancing workbook to find what needed to be balanced. The work produced a Google Doc of all known and possible reconciling items.
• Back-office automation beyond chat: The deck shows P223 enrollment automation with 2026-27 enrollment projections by school, grade and level; classified staff evaluations moved from paper to a digital, agent-guided process using open-source workflows and e-signature; and department and building budget dashboards built from the district finance warehouse.

Actionable Insights:

• Talk instead of typing: Use voice dictation to say everything you are thinking and ask the AI to help with what you want.
• Have AI write the prompt: Describe the problem you are trying to solve and ask the AI to write the best prompt for it.
• Start without knowing the answer: You do not need to know the solution before you start. The AI will offer options, as it did in the routing example.
• Build tools that support staff rather than replace their judgment: When principals were using AI to offload evaluation work, PSD built an assistant that does not evaluate teachers. It codes observations, organizes notes, flags missing elements, comments on note quality and suggests coaching conversations.

Looking Ahead:

The session's learning objectives asked business officials to identify the parts of a privacy-first governance framework, apply a good, better, best approach to prompting, and judge which of these field-tested uses could reduce staff workload in their own districts. The deck points to psd401.ai for the resources behind the talk.