01
How AI actually works.
We teach how AI actually works: tokens, context, agents, multi-agent orchestration, failure modes. Then we put it to work. Students leave able to build, not just chat.
The ground keeps moving.
Learn how to stand.
An 8-week AI training program for high school students and college graduates who want a real shot at the AI economy, without losing their judgment, their character, or their mind to it.
01 · Why now
AI changed everything in 24 months. Most schools haven't noticed.
Two years ago, most people had never used AI at work. Today, a lot of us can't do our jobs without it. Whole categories of entry-level work are being rewritten in real time. The next class of high school and college graduates will walk into a workforce their schools never prepared them for.
I'm a Managing Director at KPMG, working on how AI is reshaping the audit profession. Full time, I watch AI change how professionals actually work. Off hours, I'm building this program because my own kids are years away from being that graduate, and I couldn't find anyone teaching them what they'll actually need.
And here's the harder part:
The same tools that will define their careers are quietly eroding the skills that make them employable in the first place. Students are outsourcing their thinking before they ever learned how to do it. The result is the generation that can prompt an AI but can't write an argument, debug their own logic, or defend their own work.
We built Constare for a different kind of student. The ones who want to lead in the AI era, not get left behind by it. The ones whose parents see what's coming and refuse to hope the school system catches up in time.
02 · What we teach
Three things. Taught deeply.
02
The skills that don't expire.
Models will continuously change. The skills that matter won't. We train the durable stack: clear thinking, sharp judgment, strong communication, and adaptability. The things that compound across a career.
03
Capability with character.
Real leadership is capability and character together. We train both. Students graduate as the kind of operator others will trust with hard problems, and the kind of person who'll use that trust well.
03 · Graduate outcomes
What students walk away with
8 weeks. They ship writing. They ship workflows. They ship a working AI tool anyone can use. And they leave with a 90-day plan for what they'll do next.
-
Know the vocabulary that separates fluent from clueless. Tokens, context windows, embeddings, fine-tuning, agents, orchestration. They know what each one actually does, and when it matters for the work in front of them.
-
Build, not just chat. They've designed and shipped a working AI system that uses multiple agents handing work to each other.
-
Spot when AI is wrong. Hallucinations, bias, shallow reasoning, prompt injection. They know the failure modes and can find them on sight.
-
Think before they prompt. Every assignment starts with their own draft. AI multiplies what they bring, not the other way around.
-
Defend their own work. They can explain why every part of what they built is right, useful, and theirs.
-
Read the AI economy. They understand how AI is reshaping the workforce, where the opportunity is, and how to position themselves.
-
Lead a hybrid team. They've practiced being the human in charge of multiple AI agents: directing, reviewing, and making the call.
-
Make the call when AI can't. The hard decisions still belong to humans. They've practiced making them with conviction, not deferring to whatever the model suggested.
-
Own their voice. AI can mimic anything, including them. They keep their own writing, thinking, and point of view recognizable, even when AI is in the loop.
04 · The 8-week journey
Depth over breadth. Every week.
-
Weeks 1–2
Foundation
How AI actually works. Students leave these two weeks ahead of most working adults.
- What a model actually is: training, inference, weights, parameters
- Tokens and context: why prompts drift, why context engineering matters more than prompt engineering
- Embeddings, vector search, and retrieval-augmented generation
- Reasoning models vs. chat models, and when to use which
- The failure modes: hallucination, sycophancy, prompt injection, drift
Every lesson ends with a spot-the-error exercise. By week 2, students are catching mistakes in AI output that most adults miss.
-
Weeks 3–4
Application + Orchestration
This is where most courses stop at "prompt engineering." We go further.
- Structured prompting: role assignment, few-shot, chain-of-thought
- Agents: what makes an agent different from a chatbot, what tools they use
- Multi-agent orchestration: workflows where one agent researches, another writes, a third reviews, and the human directs
- Tool use, function calling, and MCP: how AI connects to email, databases, calendars, code
- Workflow design: when to use AI, when to use code, when to use a human
Every assignment: draft your version first, then build the AI-augmented version.
-
Weeks 5–6
Project Sprint
Students pick a problem they care about and build an AI-augmented system to solve it.
- Define the problem in plain language
- Design the solution: which agents, what tools, what handoffs
- Build it, debug it, ship it
- Document it well enough that someone else could use it
By week 6 they've shipped a working AI tool with a shareable link. And they have a written case study explaining what they built, what they found when they audited it, and what they learned.
-
Weeks 7–8
Industry + Capstone
The final two weeks zoom out.
- The AI economy briefing: what's growing, what's shrinking, what's being created
- Spotting opportunity: students identify an AI-shaped opening in an industry they care about
- Capstone presentation: students defend their decisions and explain what they'd build next
- Reflection: who are you becoming as an operator of these tools?
The capstone is the artifact they walk away with. The proof of what they can do.
05 · For students
You're already living in the AI era. Time to learn how to run it.
Your school probably isn't teaching you what AI actually is or how to use it well. The "AI class" most schools offer, if they offer one, is either an outdated Python intro or a watered-down ChatGPT walkthrough you could finish in a weekend.
Meanwhile, AI is rewriting every industry you might want to work in, start a company in, or skip college for. The students who figure this out now (really figure it out, not just play with it) will have an unfair advantage by the time they hit the workforce.
Constare exists to make that advantage real.
06 · For parents
Built for parents who see what's coming.
You've watched AI rewrite the rules in 24 months. Your student's school hasn't caught up. Constare is what you wish their school was already teaching.
What we teach
AI fluency. Not just "how to use ChatGPT." Students learn how AI actually works: the vocabulary, the architecture, the agents, the workflows.
Critical thinking in the AI era. Every assignment starts with the student's own draft. They learn to think before they prompt and defend what they built.
Character and how to hold your own under pressure. We weave judgment, restraint, and respect for people into every module. Not as a separate ethics chapter. As how the work is done.
What we won't teach
We're a small program with a clear point of view. We don't pretend to be values-neutral:
- We won't teach students to build AI companions, manipulation tools, or anything designed to exploit people.
- We don't position AI as a friend, counselor, or authority. It's a powerful coworker. That's the line.
- We won't teach shortcuts that rot the foundation. Thinking comes first. Every time.
Our foundation
We built this program around a simple idea: every kid has real capacity for judgment, character, and independent thinking. That doesn't change because AI got good. If anything, it matters more.
That belief runs underneath every lesson. It's not the daily vocabulary of the curriculum. It's the reason we built this program the way we did.
07 · About
I built Constare for my kids first.
Two of them. Both will graduate into a world where AI has impacted everything. I had to figure out how to prepare them. Constare is what I came up with.
I'm a dad to two kids. They will never know a workplace without AI. They will never apply for a job that wasn't shaped by it, never start a business without it, never sit in a college lecture untouched by it.
That should be exciting. For most parents I know, it's not. It's quietly terrifying. Because the question isn't will AI be everywhere. That's already settled. The question is whether your kid is going to be the one running it, or the one quietly replaced by it.
I couldn't find a program that took that question seriously. So I built one.
My day job is Managing Director of Technology Assurance at KPMG, where I focus on reimagining the audit through the use of AI. I spend my time building intelligence tools, deploying automation at scale across the practice, and tracking how agentic AI is reshaping the profession. I built Audit AI Pulse to track AI investments across every Big 4 and several mid-tier firms in real time. Outside of KPMG, I build web applications and tools, some for business, some to explore what's possible when you combine AI with practical problems.
So I sit at the intersection of enterprise AI and a parent's responsibility. That's the lens Constare is built on.
More on what I'm working on at marshallnunley.com.
08 · Get started
Ready to build?
Enroll once. Move at your own pace. New material unlocks each week so you're not tempted to binge and forget. Skip a week when life gets in the way, and pick up where you left off.
Get Started →
Secure checkout. You'll get instant access to Week 1 the moment payment clears.