Peninsula School District
The Open Adaptive District • psd401.ai

The Open Adaptive District — Start Here

A team of three to six staff picks one problem in their work and spends six weeks building a better way to handle it. An AI agent keeps the team’s notes. At the end, the team publishes what happened, including when it didn’t work.

We wrote this for two readers: Peninsula School District staff on a team this year, and other districts that want to run the same protocol. Everything here is free to copy and adapt. Where we name a Peninsula tool, a note says what you would need in its place.

The goals stay the same

Teams use this protocol to reach goals they already have: stronger instruction, smoother operations, better communication with families, more time with students. Wherever AI can make that work better, a team builds it in. Sometimes that means doing an old task faster. Sometimes it means changing how the task is done.

At Peninsula, those goals include commitments we have spent years building: a guaranteed and viable curriculum, our Instructional Essentials, a multi-tiered system of supports (MTSS), and Universal Design for Learning (UDL). None of them change. A team might build something that flags a student who needs support weeks earlier, or that produces scaffolds fast enough that every student gets grade-level work.

The longer aim is a district full of leaders who keep getting better at this, so that Peninsula helps decide how AI is used in schools instead of catching up later.

Why now

Children who started kindergarten in fall 2026 graduate in 2039 and will be in the middle of their careers in the 2050s. AI is already changing what it takes to be ready for work and for civic life, and new capabilities arrive every few months. Schools usually take years to change anything. If we wait for settled answers, someone else will choose them for us.

So we are doing what we ask students to do every day: try something, learn from it, and be honest about what happened. We start with adult work, such as paperwork, planning and communication, so the early mistakes don’t land on students.

Anyone can build

Building a tool used to mean writing code. Now, if you can describe a problem clearly, you can build a fix, either by asking your PSD AI Agent or in AI Studio. In AI Studio you can set up a tool, or a knowledge repository: a collection of documents, such as handbooks, procedures or past plans, that AI can search and answer questions from. A repository is a good first build. If a task eats hours of your week, try building your way out of it. The habit we want every team member to leave with is thinking “I could build something for this.”

How a cycle runs

Each cycle, a team does one build: one attempt to make part of its work better, using tools the district already has. The steps will be familiar from school improvement work, with a sharing step at the end.

  1. Plan (week 1). Pick one problem worth solving. Write a one-page build plan: what you are building, why, and how you will know it is working.
  2. Do (weeks 2–5). Build it and use it, and adjust as you go. Every Friday, the team posts four short answers in its chat space: what we tried, what we learned, where we are stuck, and what we need.
  3. Study (week 6). Look at what actually happened. The agent drafts the team’s wrap-up from the team’s own notes, and the team edits it in about twenty minutes.
  4. Share. Publish the wrap-up where everyone can see it, with one of three decisions: keep it, stop it, or run it again.

Two weeks between cycles give teams a break, and give district leaders time to read every wrap-up and set the focus for the next cycle.

Most organizations quietly drop what doesn’t work, and what they learned goes with it. We publish stopped builds alongside the ones we keep. A stopped build that saves ten other teams from the same dead end was worth running.

Teams can meet however they like: in person, in the hallway, on a call. The chat space is where the team’s record lives. After a conversation, drop in the whiteboard photo, the notes or the transcript. The agent can only use what is in the space.

When a build works

A build that works should outlast the team that made it. When a team keeps a build, we turn it into something every staff member can use, usually a new skill in the PSD AI Agent or a tool in AI Studio. The next team starts from there.

The quarterly focus

Each quarter, the superintendent writes three sentences: what we are after, why it matters now, and what we will not compromise on. Teams choose their own builds within that focus. Nobody assigns them.

The focus for this first cycle is exploration and innovation: explore the tools, and innovate on a process or practice that matches your goals.

The focus always points back to what we already value: grade-level learning for every student, MTSS, UDL, and the commitments in Peninsula 2030, our strategic plan.

Your AI agents

Every Peninsula staff member on a team has a PSD AI Agent. Each team invites two kinds of agent into its chat space: team members’ own agents, and Kris Hagel’s agent (agnt_hagelk@psd401.net), which is in every team’s space so it can report across all of them.

Neither agent touches evaluations, discipline, or anything personal. Everything they read stays on district systems.

For other districts. You need an AI assistant that can read a team’s shared chat or notes and draft from them, one agent that can see every team’s space, and somewhere to share the tools teams build. Ours are PSD AI Agents and AI Studio, both open source.

Who and when

At Peninsula, the first teams started in fall 2026: executive cabinet, principals, and district office departments such as athletics, student services, teaching and learning, and career and technical education. Each team starts when it is ready. Teacher and support-staff teams join later. Wrap-ups are shared inside the district first. Public versions, with student and staff details removed, go on psd401.ai for other districts to use.

Where to go next