Last week I did something I haven’t done before. I ran a two-day AI hackathon — not with teachers, not with tech teams at a startup, but with classified staff at Cherry Creek School District. Finance people. HR specialists. IT support. Campus office managers. The people who keep a 55,000-student district running behind the scenes.

And on Day 1, we didn’t touch a single AI tool.

Start with the Problem, Not the Tool

This is the part that surprises people when I describe the hackathon format. You’d think an AI hackathon starts with AI. It doesn’t. Or at least, ours didn’t.

The entire first day was design thinking. We mapped workflows. We identified friction — the places where people lose hours to manual processes, repetitive questions, and tasks that feel like they should have been automated years ago. We asked participants to get specific about their pain points. Not “AI could help with stuff.” More like “I spend five hours a week answering the same benefits questions by email.”

That specificity matters. Because the biggest failure mode in AI adoption isn’t the technology — it’s implementing solutions to problems nobody actually has. When you start with the tool, you end up looking for nails. When you start with the pain, you build something people will actually use.

By the end of Day 1, every team had a clearly defined problem, a target user, and a vision for what a solution could look like. No code. No prompts. Just clarity.

Day 2: From Consumers to Builders

Day 2 was where it got real. Teams took those well-defined problems and built working prototypes using Google AI Studio. In a single day.

Here’s a sample of what they created:

One HR team built an AI assistant trained on the employee handbook and payroll calendar. Staff can now ask plain-language questions and get sourced answers instantly — and when the answer isn’t in the handbook, it tells you to contact HR directly instead of making something up. They estimated it could save over 200 hours per year across 40 office managers.

An IT team used Salesforce AI to auto-triage incoming support cases, pulling in missing information and routing tickets to the right queue. Their estimate: 8+ hours saved per week for IS staff, with faster resolution times across the board.

A finance team built a bookkeeping tool to simplify financial adjustments — turning a clunky, manual process into something their team actually wants to use.

Another team tackled a problem every large organization knows well: nobody knows what software they already have access to. They built an AI-powered directory that guides staff toward existing tools instead of requesting new subscriptions — reducing duplicate costs and cutting through red tape.

A performing arts team created a single source of truth for districtwide event data, pulling from a Firebase database into calendars, flyers, and website updates — all from one simple input interface.

A grants team built a tool that reads through dense grant award documents, extracts the key information, and structures it into tables ready for export. What used to take hours of reading now takes minutes.

And one team started building a Python script to automate PDF transfers between vendor platforms and their student information system — a task that was eating up staff time every single week.

The Numbers Tell the Story

Before the hackathon, participants rated their AI confidence at 2.62 out of 5. After two days, that jumped to 4.00 — a shift of nearly 1.4 points. One participant described it as getting training wheels on a bike they’d been afraid to ride.

Pause on that for a moment. These aren’t software engineers or AI researchers. These are operational staff — people whose primary job is finance, HR, IT support, office management. And 94% of them said their hackathon solution has real implementation potential. Half said it only needs minor refinement before it’s ready for pilot.

The overall experience was rated 4.75 out of 5. When asked whether they’d recommend this to a colleague, the average score was 9.62 out of 10. When asked what they’d tell a colleague beforehand, the most common answers were simple: “Come with a specific problem” and “Just do it.”

This Is an Operations Story, Not Just an Education Story

I want to be direct about something. This hackathon happened in a school district, but the problems these teams solved are not unique to education. Every organization has an HR department drowning in repetitive questions. Every IT team has a support queue full of tickets missing basic information. Every finance team has manual processes that could be streamlined. Every grants or procurement team spends hours reading documents that AI could parse in seconds.

The model works because it starts with people and their actual work — not with technology looking for a use case. The design thinking on Day 1 is what makes the builds on Day 2 meaningful. And the psychological safety of the hackathon format — the permission to experiment, to fail, to try something you’ve never done — is what unlocks the shift from “AI is intimidating” to “AI is something I can actually use.”

That shift is what matters most. Not the prototypes themselves — though many of them are genuinely impressive — but the moment when someone realizes that AI isn’t just an answer machine. It’s a thought partner. And that they’re capable of building with it, not just consuming it.

What Makes This Work

A few things I’ve learned from running these sessions that I think apply whether you’re in a school district, a nonprofit, or a corporate team:

Don’t skip the problem definition. It’s tempting to jump straight to tools. Resist that. The quality of what gets built on Day 2 is directly proportional to the clarity achieved on Day 1.

Make it safe to be a beginner. Most of our participants had never built anything with AI before. The hackathon format gave them permission to try — and that permission is more powerful than any tutorial.

Celebrate real work. These weren’t toy projects or hypothetical exercises. Every team worked on a genuine problem from their actual job. That’s what made the showcase at the end so energizing — people could see their own frustrations being addressed by their own colleagues.

Keep going after the event. The hackathon is a starting point, not an endpoint. Several teams are already planning pilots, and the district is exploring how to support implementation of the most promising solutions.

Bring This to Your Team

If you’re leading an operational team — in education, government, nonprofit, or the private sector — and you’re wondering how to move your people from passively using AI to actively building with it, this is the format. Two days. Real problems. Working prototypes. A team that walks out more confident and more capable than when they walked in.

At UnconstrainED, this is the work we do. We help organizations design and run AI hackathons, professional learning experiences, and implementation strategies that meet people where they are — not where the hype cycle says they should be.

If you’re interested in bringing this to your organization, reach out at alex@unconstrained.work. I’d love to hear what problems your team is sitting on.

— Alex