How long is your to-do list? What great ideas for your district would you explore if only you could find the time to escape the routine tasks?
Many school business leaders know that AI can address this time gap, and yet, according to a recent Gallup Poll, only 10% of the workforce uses AI daily. Plenty of us have experimented by asking ChatGPT to solve our immediate needs: Find the best flight to Denver, provide the best way to remove age stains in clothing, suggest a birthday gift for Mom. As Ross Hartmann, the founder and CEO of Kiingo AI, knows, this isn’t a technology problem.

“It’s a leadership vacuum,” he says. “When leadership is absent from the conversation, employees fill that void themselves – personal accounts, no guardrails, security risks invisible to the organization. Everyone has the same AI tools. The winners will be [those] whose managers actually lead their teams into using them.”

Hartmann applied his hands-on engineering and architecture roles at ProMAX Systems and LifeScore Music (he attended the Berklee College of Music for songwriting in 2008) to establish his own AI consultant company in 2023, and since then has trained more than 4,000 executives across 600 companies. He helped CASBO develop its AI programs.

His main focus: Showing leaders how to find that critical time eluding the workforce. He’s personally passionate about how his firm now handles incoming client email. Ask AI to draft a reply?

That’s a vanilla start. He set up a system to read the whole chain, combine it with meeting notes from the client, and draft three responses to choose from. He’s also big on having Claude record a meeting, paste the transcript, ask for a summary + action items with owners – and take 90 seconds rather than a 20-minute recap. Want to try this yourself? Paste your last 20 sent emails, ask AI to find patterns and build response templates. Most teams, he says, have eight to 10 email types that cover 80% of outbound correspondence.

Messy Excel file with inconsistent headers and too many tabs? Ask AI “What should I know about this before Monday’s meeting?” Boom. A clean executive summary in less than 60 seconds.

Or try this: Upload meeting notes and project docs and ask AI to organize everything into a structured presentation. It turns a two-hour, blank-slide problem into a 15-minute starting point. Hartmann estimates these little tasks can save 10 hours per employee a month.

“Most teams aren’t short on AI access. They’re short on results from it,” he notes.

CASBO sat down with Hartmann to discuss key aspects of integrating AI into school business.

What’s the best advice you’ve ever received? From whom?
Charlie Munger had this idea he borrowed from a mathematician named Jacobi: “Invert, always invert.” Instead of asking, “How do I succeed?” you flip it and ask, “What would guarantee failure?” Then you just avoid those things. I use this all the time, both personally and when I’m working with leadership teams on AI adoption. Instead of asking “How do we roll out AI successfully?” I’ll ask “What would make this rollout fail?” And the answers are usually pretty obvious: no training, no clear policy, no leadership buy-in, expecting people to figure it out on their own. Once you invert the question, the path forward gets a lot clearer. You’re not guessing at what might work. You’re just making sure you’re not doing the things that are almost guaranteed to fail.

As someone who trains leaders, you’ve seen fear of the unknown firsthand. What’s something you initially believed to be a permanent barrier to organizationwide AI adoption that you now know is a solvable leadership problem?
Two things: the AI Reflex and AI Conviction. The AI Reflex is how often you actually turn to AI when you hit a problem. It’s your instinct in the moment. Do I think to use AI to help me think through this, pressure test my idea or actually do the work for me? Even people who know AI can do those things don’t always reach for it in the moment. But it’s a muscle. You can develop it. AI Conviction is a little more nuanced. It’s the belief that when you hit a technical hurdle, when the AI doesn’t understand you, when the output isn’t great on the first try, you have the conviction that it will work and that these tools are getting better constantly. You just have to sit down and clarify your own thinking so the AI can actually help you.

It’s the old Henry Ford line: Whether you think you can or you think you can’t, you’re right. I tend to find that people who believe AI can’t do something won’t push past the initial friction. And the people who believe it can? They push through, and they see materially better results.

You’ve said that a leadership vacuum leads to employees using personal accounts without guardrails. What are three non- negotiable policies school business leaders should implement to move from accidental AI use to a secure district standard?
This is a big one. In technology, we call this shadow IT, which is where someone on your team wants to use a new platform, but they know that if they ask the IT team, the answer is going to be, “Absolutely not, we have to vet this vendor, we have to spend six months looking into it,” and so on. So, what does that person do? They use it anyway, just under the table. We’re already seeing studies that show the same thing happening with AI. It’s called shadow AI.

The three nonnegotiable policies:
First, have a clear AI use policy so that people know what your local education agency actually allows and what it disallows. Without that clarity, people either don’t use it at all, or they use it in ways you can’t control.
Second, have your IT team set up the infrastructure properly so that data privacy and security come first. This means managing the connectors between your AI tools and the software you already use, configuring your AI settings so the models don’t train on your data, and so on. Third, training, so that people know how to use the tools properly. When you combine a clear policy, secure infrastructure and solid training, your team can use these tools without putting your district at risk.

How does a superintendent shift the culture from “AI as a toy” to “AI as a core administrative utility?”
It’s two things: training and experience. You need the training so that you understand how to use the tool on a foundational level and you can get over the initial friction of using a new technology. But you need the experience to really understand where these tools apply to your specific workflows, and that only comes from time and practice.

There’s no magic here. If you have a solid foundation and you’re actively trying to apply AI to your workflows, it will naturally move from “toy” to “core utility.” But experimentation, and being willing to work past that initial friction, is the whole game. The leaders who treat it like a novelty will get novelty results.

You’ve said teams aren’t short on AI access, they’re short on results. What is the most common mistake leadership teams make during the first 90 days of an AI rollout that prevents them from seeing a real ROI?
There are a number of mistakes, but the biggest one is that leaders will just roll out AI and expect everyone to immediately be more efficient. But if I’m using a technology for the first time and I haven’t really understood how it works on a foundational level, I’m going to use it incorrectly, or at least not to its full potential. And that causes me to think these tools aren’t very good, which develops into cynicism, maybe even skepticism. You’re already starting behind if that’s where people land.

This is why training matters so much at the beginning. The second mistake is the type of training. A lot of leaders will run a two-hour workshop that gets people excited and motivated, but when those people get back to their computers to actually use AI, they find they don’t have the core click-and-type skills to make the technology work. And then there’s the issue of not providing a clear AI-use policy. If there’s no direction, people aren’t going to have the confidence to use the tools within your organization. You’re going to see more shadow AI, not less.

You estimate AI can save 10 hours per month per employee. In a district office setting where staff is juggling dozens of tasks and requirements, where do you see those 10 hours being recovered first?
When AI is used as an augmentation tool, it really is more about saving 15 minutes here and 15 minutes there, rather than saving five hours on any one task. There’s no single area where all 10 hours come from.
It’s things like responding to email faster, with more confidence, or with less back-and-forth. As an example, I have an AI agent that drafts responses to my emails every hour. It knows how to do that because I’ve trained it, and it can research the participants on an email chain, pull up past conversations, past meetings and relevant context before I even open the email.

Beyond email, you recover hours from things like document summarization, creating memos, analyzing reports and processing invoices. Asana calls this “work about work,” which is the time you spend looking things up, formatting data, copying information from one app to another. There’s a surprisingly large chunk of anyone’s week that falls into this category, and you can recover a lot of it with the right AI tools.

You noted that AI can turn two-hour presentation prep into a 15-minute starting point. How can school leaders use this specifically for high-stakes communication like explaining complex topics to school board members or the public?
There are really two ways to approach this. The first is to just let the model have a first pass at the communication. Give it the complex topic and ask it to explain it for a school board or public audience. The model will likely do pretty well right out of the gate. But the better method is to take some time and think about context.

Do you already have writings where you’ve explained this topic before? Has someone on your team? Can you feed that existing material into the model as a starting point? AI does well on its own, but it does much better when you combine it with your own judgment and context.

The other thing I’d say is: Use the smartest model available. Don’t use the free version of any of these platforms. Don’t use the default model. In Claude, that’s their Opus model. In ChatGPT, that’s their Pro model. The best model is going to give you the best answer, especially for high-stakes communication where nuance matters.

School districts are notorious for having data siloed across inconsistent spreadsheets. How can a chief business official use AI to bridge the gap between “data-heavy” and “decision-ready” before a cabinet meeting?
This is where AI agents come in, because agents can be given secure access to your district’s data. Instead of having someone on your team manually upload spreadsheets to an AI tool, you give the AI direct access to your spreadsheets, wherever they live, whether that’s SharePoint or Google Drive. You do this securely using what are called connectors, or MCPs (model context protocol). Once that’s in place, the AI can pull and analyze data on its own, which cuts out a lot of the time wasted on finding, formatting, and consolidating information.

At that point, as a chief business official, you can just start asking for what you need. The AI already has access to the data and can help you get to the answer, whether that’s a summary, a comparison or a trend analysis you can walk into a cabinet meeting with.

You’ve said that the “winners” will be those whose managers actually lead their teams into using AI. Five years from now, what will a “winning” California school district look like compared to one that remained absent from the conversation?
In five years, the winning California school district will have cut a huge amount of administrative work so that more time goes toward actually serving students and faculty. But they’ll also be doing things that just weren’t possible before because of the time they would have taken.

Think about personalized learning plans that adapt to each student, not because a teacher suddenly has 30 extra hours a week, but because AI is helping them understand where each student is and what they need. Think about teachers getting real support on lesson planning, grading and parent communication, so they can focus on the parts of teaching that actually require a human. Think about district leaders running
budget scenario analyses in minutes instead of weeks, or tracking legislative changes and immediately understanding how they impact the district.

The districts that sit this out aren’t just going to be behind. They’re going to be spending more time and more money to get worse outcomes. And every year the gap gets wider.