top of page

An AI Framework for Schools: Examples From Franklin, Miami-Dade and Seckinger

Nov 5
13 min read

This AI framework for schools is built from three examples that took three different paths. Franklin School is a small private high school in Jersey City that won the 2025 World's Best School Prize for Innovation. Miami-Dade County Public Schools is a large public district that moved from a ban to a staged pilot. Seckinger High School in Georgia is a public school built around AI since 2022. Here is what school leaders can borrow from each, starting with Franklin's Will Campbell.


In this article


  • Try it first: the Teacher Time Test. A free two-week check on whether an AI tool gives one teacher time back. No gatekeeping, it's below. Jump to the tool.

  • The framework: six steps, built from how Franklin School rolled out AI. The steps are my organization of what Will Campbell described.

  • The sandbox came first. Franklin tested AI three or four times before rolling it out to students and staff.

  • Teacher time came first, too. Teachers heard about 8 to 9 hours a week back before they heard anything about the tools.

  • Harder, not easier. Campbell wants AI to make school harder where it counts, with tougher assessments and deeper questions.

  • Two more examples: Miami-Dade County Public Schools (a district) and Seckinger High School in Georgia.

  • A 30-day plan you can start next week, plus an FAQ.



The challenge: AI arrived before the guidance did


Principals and district leaders are being asked to make decisions about AI before the rules are clear. RAND's national surveys found that in 2025, 54% of students and 53% of English, math and science teachers used AI for school. Only 45% of principals said their school or district had AI policies or guidance. Just 34% of teachers said they had a policy on AI and cheating. And more than 80% of students said their teachers had not explicitly taught them how to use AI for schoolwork.


Leaders are also hearing mixed messages. In the same RAND report, 61% of parents agreed that more AI use will harm students' critical-thinking skills. Only 22% of district leaders agreed. And in an EdWeek Research Center survey fielded from August 26 to October 6, 2026, nearly one in three district and school leaders said they don't use generative AI for any operational purpose.


On paper, that looks like a technology problem. It reads more like a sequencing problem: who tests, who trains, who decides, and in what order. A few schools have answered those questions in public. They are worth studying. For my earlier take on AI in schools, see AI in Education: A Tool, Not a Threat.


Three examples worth studying


  • Franklin School (Jersey City, New Jersey): a private high school that, according to its founding head, tested AI in a sandbox three or four times before rolling it out. It also protected teacher time before asking anything of teachers.

  • Miami-Dade County Public Schools (Florida): a public district that blocked chatbots in 2023. It then spent months testing tools, trained more than 1,000 educators, and partnered with Google on a staged rollout of Gemini to more than 105,000 high schoolers, according to WLRN.

  • Seckinger High School (Buford, Georgia): a Gwinnett County public high school that opened in fall 2022. It calls itself the nation's first AI-themed high school and builds AI into instruction across subjects, according to the Atlanta Journal-Constitution.


I run Young CEO Squad, and our non-AI workshops are used in over 400 classrooms. So I pay close attention to how schools decide what to let in. We have also been training heavily in AI so we can build CTE curriculum and courses for schools. Our goal is for students to learn to build the tools that today's small main-street businesses need. Not only do we want students to leave high school able to work as an AI consultant, we can even teach them younger.


From the teacher responses and conversations we have seen, teachers adapt materials to fit how they teach, the time available, and where their students are right now. Sometimes a rollout prescribes the implementation in rigid steps. That is fine for the implementation phase. It is not fine for getting teachers to adopt a new program. They need room to put their own take on it. They are the experts on what will work in their classroom.


As I started listening to the AI in Education podcast I was immediately drawn to the question of how can my children's schools become a leader in innovation like Franklin. I went through Will Campbell's full conversation with Dan Bowen and Ray Fleming on The AI in Education Podcast, then checked what I could against public sources. Where Campbell is the only source, I say so.


When most schools were debating whether to ban AI, Franklin was on its third round of using it. Campbell says his team had tested the tools in a sandbox three or four times before the ban conversation got going. Franklin has about 182 students by one directory's count. That is a very different scale than a district. And yet the order he describes holds up at any size.


Campbell did not present his approach as a framework. I organized what he said into six steps, and that framework is mine, not a published Franklin model.


Listen to the full episode: Building Superhumans: Reinventing Schools for the AI Age, from The AI in Education Podcast, hosted by Dan Bowen and Ray Fleming. Published October 15, 2026.







An AI framework for schools, in six steps


Each step starts with what Franklin did, then shows how a school or district of any size could use it.


1. Start in a sandbox


Franklin’s answer to AI was to test it inside the school first, then let it out in small pieces. The team ran it through a sandbox three or four times to see what might happen if AI became part of everyday school life. After that, they brought it to a couple of student groups and built a short course on AI and machine learning. A director of innovation, who is also an associate professor at NYU, served as a thinking partner.


Franklin's tagline is High School Reimagined. Campbell says the school told its community up front that there would be problems and that the rollout would be gradual. While other schools put up fences, his team built the systems to manage the risks.


A sandbox gives room for improvement and recovering from surprises. If your first real test of a tool happens in front of 30 students and a skeptical parent, you have already lost the argument.


Pick one AI tool (a chatbot, for example), one volunteer teacher and one unit. Write down what goes wrong before you tell anyone else. That is my suggestion, not Campbell's, but it follows the same logic.


2. Give teachers their time back first


The conversation with teachers did not start with the tool. Campbell says the school told them AI would not replace them or take over their thinking. Then it made a promise: AI could give them 8 to 9 hours back each week. Time, he says, is where the conversation started. He described the 8 to 9 hours as a promise made to teachers, not a measured result. Our proposed way to measure time saved is the Teacher Time Test, below.


Franklin put resources behind that promise. Faculty and staff have had licenses for a long time. Professional development takes two periods a week and counts toward the teaching load. There is also Strategic Saturday, open to anyone who wants a hand in shaping how the tools get used.


Campbell says fear of job loss has dropped a lot, but it still comes up. It comes up mostly with people who dislike AI tools. He also says teacher retention is strong, from the newest teachers to the most experienced. He gave no numbers, so I am not adding any.


The part I would borrow is where the professional development sits on the schedule. At Franklin it counts toward teaching load, so it is not homework. You may not be able to rebuild your master schedule overnight. Start with one protected, recurring block.


3. Treat AI as a sparring partner


Campbell described how he uses it himself. He asks AI for a snapshot of a large body of research. He checks the snapshot, then reads the specific papers on his own. It saves him time. He says the tool became a thinking and sparring partner, not an assistant doing work he should be doing.


We've spoken on that topic and agree. The tool can summarize. The person still has to question the summary. It is also a good test for any policy you write: is the tool doing the thinking or sharpening it?


4. Learn from students before you scale


Campbell says students helped work out how to use AI before the school did. The short courses became an elective. Franklin Student Ventures, a Shark Tank-style incubator, gave students a place to build with the tools. Students whose parents worked in the field also fed into the effort.


I wondered whether this meant a student committee or formal proposals. Campbell did not describe either one, so I will not invent a process for him. What he did describe is students trying the tools and shaping what the school did next.


5. Make school harder where it counts

AI is going to "make the school experience harder where it counts."

Most of what I read about AI in schools promises easier. Campbell's version asks more of students. If an AI tool can finish an assessment quickly, he says, the assessment is not hard enough.


Franklin's skills curriculum puts weight on judgment and ethics. He sees these as the skills that set people apart as AI speeds up knowledge.


Here is what that means for a district. Take three take-home assignments from last semester and ask whether a free chatbot could finish each one in a few minutes. For most written and problem-set work, the answer will be yes.


Where the only evidence is a finished document, add a step where the student shows their thinking in person. They could explain it aloud, work it in class, or use results from a real project.


I would add one thing from my own work. A policy cannot sit next to a student when an assignment gets hard and the shortcut is one tap away. A student who has decided for themselves that the shortcut will not help can. So the raise comes second. First you meet students where they are, then you build them up.


Harder only counts if students have somewhere to stand when it arrives.


6. Bring in outside partners


Franklin did not do this alone. In its first year the school became a professional development hub. Campbell says it is also the leading fab node for North America, and it invites educators into its fab lab. He has also worked with Hewlett Packard on two AI reports over three or four years. His own summary is that schools need outside partners to make this work.


Beyond Franklin: Miami-Dade and Seckinger


Miami-Dade shows a common backstory. The district blocked chatbots in 2023 over fears of cheating and misinformation. It then tested nearly a dozen tools for accuracy, privacy and fairness, and staff posed as teen hackers to look for harmful outputs. It chose Gemini partly for its teen-specific guardrails. It gave live demos to nearly 400 local principals and launched an AI Institute for teachers. The student pilot started in spring 2025 at Southwest Miami Senior High. Google, Miami-Dade's partner on the Gemini rollout, says in its own announcement that the program reaches 100,000 high school students and that all 18,000 teachers will take its Generative AI for Educators course. It also lists two planned pilots, an AI tutor and a tool that helps administrators read data.


WLRN also reports the limits. It cites a RAND finding that the likelihood generative AI tools are leading to measurable improvements in teaching and learning is low. The assistant superintendent who designed the program said he hoped it would transform learning. The district has no formal outcome data yet. The sequence can be helpful: test, train, demo to principals, then pilot.


Seckinger High School took a different route. It did not start with a pilot. Gwinnett County Public Schools began building a cluster of AI-focused schools in 2019 and opened Seckinger in fall 2022, about three months before ChatGPT launched, according to IBM Think. The district chose not to ban ChatGPT and treated it as a tool.


Teachers are expected to build one or more of six components into their lessons: programming, data science, mathematical reasoning, creative problem solving, ethics and applied experience. They must also teach the ethical use of any AI tool for at least a week before students get it. Principal Jimmy Fisher told IBM Think, "We want our students to walk away being proficient users, but also ethical users." Students can take a three-course AI career pathway, as the AJC reported in 2025.


The district now pays for Microsoft Copilot and Google Gemini. Social studies teacher Scott Gaffney says his students finish a data analysis in one class period instead of several days, which leaves more time for discussion.


7. Start earlier: the mindset AI rewards


Franklin starts at ninth grade. I would start the groundwork years before that. The habits that make a student good with AI are not AI skills. They are a willingness to try something with no guarantee, to hit a roadblock and adjust, and to stay with a problem while the answer is unclear. Campbell touches the same ground when he talks about helping students sit in the unknown and understand that not knowing is okay. For more on how entrepreneurship and AI skills connect, see More Crucial Than Ever: Entrepreneurship and AI Skills.


A classroom-based small business is a low-stakes place to practice exactly that. Our 3rd to 5th grade guidebooks walk students through setting a goal, treating obstacles as creative challenges, pricing, and asking for the sale. Picture a student whose price turns out to be too high. Nobody hands over the answer. The student changes the price and tries the next customer. For the broader case for starting early, see The Importance of Entrepreneurship Education.


That is why our hands-on workshops begin in 3rd grade, with the lesson plan and materials handled for the teacher. It is also why our Business in a Box workshop model is where many classes start. The work grows over the grades. For teens it grows into AI School for Teens, an online course we are building now for ages 13 to 18. Students learn how to diagnose business challenges that AI can solve and then build the solution. By then they have already built and run something small. So the question they ask is "what can I build with this" and not "will this take my job."


I do not have outcome data showing that early entrepreneurship produces stronger AI users. That is my belief, drawn from how we built our programs.


AI School for Teens is not open yet. Join the waitlist and you will be first to hear when it is.


What to be careful about


Four limits apply to everything above.


First, most of the Franklin detail comes from Campbell himself in one podcast conversation. I checked the facts I could against Franklin's website, T4 Education, and school directories. I could not independently confirm his claims about teacher time saved, retention, or the fab node.


Second, the prize recognizes innovation. It is not a measure of test scores or learning outcomes. Franklin has not published outcome data that would let anyone compare it with other schools.


Third, the national survey figures come from 2025 and 2026 and will keep moving. Fourth, Franklin is a private school of roughly 182 students. A public district with many schools cannot copy these moves one for one. The order of steps transfers. Franklin's specific arrangements, like two professional development periods a week counted in the teaching load, may not.


How to roll out AI in schools: the next 30 days


  1. Week 1: Choose one AI tool, such as a single chatbot, and one volunteer teacher. Have the teacher log how long a few repeat tasks take, such as lesson planning, feedback and parent emails. Then test only that tool, privately, and write down what breaks.

  2. Weeks 2 and 3: Have the volunteer repeat the same tasks with the tool. Share the difference in hours with other teachers before you tell them what they will need to do differently. Reserve one recurring block of time for training.

  3. Week 4: Audit three assessments. For any a chatbot could finish in minutes, add a step where the student shows their thinking in person.


One teacher over one week gives you a rough estimate, not a guarantee. Share it as one colleague's result.


Get the Teacher Time Test



Bring a hands-on workshop to your school


If you want to start with students instead of software, a Young CEO Squad class workshop is a turnkey way in. See a sample lesson plan or choose a workshop.


Frequently Asked Questions


What is an AI framework for schools?

An AI framework for schools is a plan for who tests AI tools, who trains staff, who sets the rules, and in what order. The six steps in this post come from Franklin School's rollout, as described by its head of school, Will Campbell. They are my synthesis, not a framework Franklin has published.


How should a school start rolling out AI?

Pick one AI tool, one volunteer teacher and one unit. Test it privately, write down what goes wrong, then protect time for teacher training.


Is Franklin School public or private?

Franklin School is a private, independent, co-ed high school for grades 9 to 12 in Jersey City, New Jersey. Franklin's website says it was founded in 2021.


What award did Franklin School win?

In 2025, T4 Education named Franklin School the winner of the World's Best School Prize for Innovation.


Who is Will Campbell?

Will Campbell is Franklin School's Founding Head of School. According to Franklin's website, he chairs the Franklin Foundation and sits on the Dwight International Heads Council.


How did Franklin introduce AI to teachers?

Campbell says the school told teachers AI would not replace them and promised 8 to 9 hours back each week. Faculty and staff also have licenses, and two periods a week of professional development count toward teaching load.


How can an elementary or middle school use Franklin's approach?

Test tools privately, protect teacher time and give students a hands-on project. Young CEO Squad's hands-on workshops start in 3rd grade. AI School for Teens, in development for ages 12 to 15, picks up in the teen years.



About the Author Suzanne Appel is the founder of Young CEO Squad, a hands-on entrepreneurship and financial literacy program now used in over 400 classrooms. She launched YCS in 2019, inspired by her own kids' enthusiasm for selling, and brought 25 years of corporate digital marketing experience to the mission of making entrepreneurship accessible to every child, regardless of means. She and her husband Scott are raising two kids in Agoura Hills, CA, a down-to-earth community north of Los Angeles with great schools and lots of wild rabbits.


 
 
 

Comments


Kids Business Start Up

Empowering kids in grades 3-12 with hands-on entrepreneurship education that builds confidence, creativity, and real-world skills.

Kids Entrepreneur

Shop

Company

Groups/ Classroom Products

Mystery Pack Workshop

Keychain Start-Up Kit

Vinyl Sticker Kit

Reusable Straw Kit

Refill Packs​​

Classroom Posters​

About Us

Contact

805-242-2005

info@youngceosquad.com

Empowering the next generation of entrepreneurs since 2019

  • Facebook
  • Instagram
  • TikTok
  • Pinterest

© Young CEO Squad, LLC

bottom of page