🧭 Fundamentals
The mindset shift and the vocabulary, in plain language. Without being technical, you’ll understand what an agent, a skill, and AI memory are—and why, even with all of that, you’re still in charge.
0 of 30 topics
Look at the rungs: each module takes you up one level — from user that only asks conductor that’s in charge.
Course roadmap
Detailed content
🎼 The turning point: from user to conductor
What changes when AI stops being a chat and becomes an executor — and why this is the best time to get started.
We used to use AI to get answers to questions. Now it handles entire tasks—writing, organizing, programming, researching—while you supervise.
Anyone who understands this shift stops “asking for little bits of text” and starts delivering real results.
Using vs. commanding · AI as executor · human oversight.
Making an app, writing a text, or creating an automation costs almost nothing today. So value has shifted from “knowing how to do it” to “knowing what’s worth doing.”
That’s what separates people fighting for scraps from those who charge a premium: judgment, not typing.
Barrier falls · value shifts · judgment.
AI changed phases almost every year: chatbot → automation → agent → agentic (AI that goes and gets things done). We’re in the agentic phase.
Those who get in early at each stage catch the wave; those who stay in the old stage compete in a crowded market.
Waves · get in early · agentic phase.
The conductor doesn’t play an instrument: they conduct. Your new role is to conduct the AIs — tell them what to do, when to do it, and check the result.
It's the mental image that guides the entire course.
Govern · delegate · check.
Tools change every month. What doesn’t change is knowing a real problem and understanding what matters. You already have that.
Take the fear out of “I’m not technical” — your experience is exactly what sets you apart.
Skill > tool · experience · problem context.
Direct AI to build real things and set up a personal assistant (Jarvis) that works for you.
Knowing where you’re headed keeps you motivated along the way.
End result · your Jarvis · portfolio of proof.
🤖 Anatomy of an agent
What an AI “agent” is, demystified: how it thinks, acts, and why it tries again on its own.
A chatbot answers. An agent gets a goal and carries out the steps on its own until it's done.
Knowing the difference keeps you from asking too little of a tool that can do a lot.
Answer vs. act · goal · autonomy.
A simple cycle: look at the situation, decide the next step, do it, and repeat.
Understanding the cycle helps you give better instructions.
Cycle · step by step · iteration.
These are actions the agent performs: search the web, open a file, run a command.
Without tools, AI only talks; with them, it acts.
Action · tool · capability.
If one step fails, the agent notices and tries another approach without you asking.
That’s what makes it feel like “it’s working.”
Loop · self-correction · persistence.
It can "make things up" or confidently go down the wrong path. You need to check.
The conductor hears the wrong note — that’s your job.
Hallucination · review · fact-checking.
Automation follows a fixed path. An agent adapts when the situation changes.
You choose the right tool for each task.
Rigid vs. adaptable · choice.
🔌 Skills, MCP, and the Ecosystem
The pieces that turn standalone AI into a system that does things in the real world.
A skill is a ready-made recipe that teaches AI to do a task your way, the same way every time.
It’s how you package your knowledge so AI can repeat it.
Recipe · pattern · reuse.
MCP is a standard "plug" that connects AI to services (email, spreadsheet, calendar) without hacky workarounds.
That’s what connects your Jarvis to the world.
Pattern · connector · integration.
Web, files, apps, databases—each connector gives AI a new capability.
You build the toolkit based on what you need.
Powers · toolkit · combination.
A growing world of ready-made skills and connectors created by the community.
You don't need to build a lot of things — just plug them in.
Community · ready-made · plug in.
When memory, skills, and connectors come together, AI becomes a hub that runs everything—like a computer’s OS.
It's the foundation of the Jarvis you build in Track 3.
Hub · layer · Jarvis.
Start small: one skill + one connector that solve a pain point for you.
Keeps you from getting stuck trying to build everything at once.
Start small · one pain point · incremental.
🧠 Context, memory, and the harness
Why AI sometimes “forgets”—and how to keep your agent focused and on track.
Everything the AI has “in its head” during that conversation: what you said and what it has already done.
Good context = good answers. Messy context = confusion.
Short-term memory · conversation · focus.
The AI’s “head” has limited capacity. A very long conversation pushes the beginning out.
Explains the “forgetting” and teaches you how to work around it.
Limit · window · summary.
Notes the AI keeps to remember your preferences between conversations.
That’s what makes Jarvis “know” you.
Persistence · preferences · continuity.
It's the structure around the AI that organizes tools, memory, and rules—the “car” where the engine (AI) is installed.
A good harness makes an ordinary AI perform much better.
Structure · chassis · organization.
A clear goal, small steps, and checks keep the agent from getting tangled up.
Long tasks require this kind of guidance.
Goal · steps · checks.
One conversation per task, give examples, and start fresh when things get tangled.
Small habits greatly improve the result.
Habits · examples · fresh start.
🩺 The human advantage: diagnose > build
The part no one outsources to AI: finding the right problem and proving the solution worked.
The pharmacist delivers what you ask for. The doctor figures out what you need. The doctor earns more.
Your value lies in diagnosing, not just executing.
Diagnosis · need · value.
Instead of automating just anything, find the bottleneck that, once solved, unlocks the result.
Solving the right bottleneck is what matters — everything else is wasted effort.
Bottleneck · constraint · priority.
A KPI is just a target number: time saved, sales, fewer errors. You choose it before building.
Without a number, no one sees the value of what you did.
Target number · measure · before you start.
After building, testing, and measuring whether the number actually moved.
"It looks like it worked" isn’t proof. Measuring is.
Test · measure · evidence.
In the AI market, showing what you've done is worth more than a certificate. Keep proof.
"What have you already built?" is the question that decides it.
Portfolio · proof · build in public.
In a few years, "knowing AI" will be a basic skill for everyone — like every accountant uses Excel today.
The advantage is real, but the window won’t stay open forever.
Window · early · temporary advantage.