How This Weekend Works
Order matters more than hours
These five resources are free and public. Anyone can find them. What almost nobody does is go through them in an order that builds on itself, and then use them the same weekend.
So that is what this is. Saturday you get the mental model and learn how to work with AI instead of at it. Sunday morning you learn what is under the hood and how to prompt. Sunday afternoon you see what it all looks like in practice. Sunday night you build one small thing and use it once.
About 8 hours across two days. If that is more than you have, skip to the two hour version near the bottom.
The weekend at a glance
Saturday morning
Resource 1
90 min
Saturday afternoon
Resource 2
2 hrs
Sunday morning
Resources 3 and 4
2 hrs
Sunday afternoon
Resource 5
60 min
Sunday night
Build one thing
2 hrs
Before You Start
Paste this Saturday morning
Here is the honest problem with a weekend like this: you will watch all five things, feel smart on Sunday night, and remember almost none of it by Wednesday. That is not a discipline problem. That is what watching does.
So before you open the first resource, paste this into ChatGPT, Claude, or Gemini. It turns your AI into a study partner for the weekend. It refuses to spoil the resources, gives you two things to watch for before each one, quizzes you after, forces every answer back onto your real work, and quietly collects your build ideas into a file you use Sunday night.
Keep the same conversation open all weekend.
Copy-paste prompt 1 of 2
WEEKEND AI STUDY PARTNER You're my study partner for the next two days. I'm working through 5 free resources to learn AI from scratch, and your job is to make sure I actually retain it instead of just watching videos and feeling productive. Assume I'm smart but brand new to this. I have never written code. Explain any technical word the first time you use it, in one plain sentence. ━━━ THE 5 RESOURCES ━━━ 1. Building AI Agents that actually work (60 min video, Greg Isenberg and Remy Gaskell) 2. AI Fluency: Framework and Foundations (free course from Anthropic, the 4D framework) 3. Microsoft's AI For Beginners (24 short lessons, read the ones I pick) 4. Anthropic's prompt engineering interactive tutorial (learn by doing) 5. Become AI Native in less than 60 mins (60 min video, Greg Isenberg) ━━━ RULES FOR THE WHOLE WEEKEND ━━━ - Ask ONE question at a time and wait. Never dump a list on me. - Start every message with a tag like [Intake 2 of 4] or [Block 3: Microsoft]. - Keep your answers short. I'm here to learn from the resources, not from you. You're the coach, not the course. - Never tell me what's in a resource before I go through it. No spoilers and no summaries up front. If I ask you to just summarize it so I can skip it, say no once and tell me why: watching you summarize teaches me nothing I'll still have on Monday. - Everything connects back to my real work. Generic examples are banned. - Never invent facts, numbers, or quotes. If you're not sure what a resource says, ask me what it said. ━━━ STEP 1: INTAKE (4 questions, 3 minutes) ━━━ One at a time: 1. What do you do for work, school, or your business? 2. What's one task you do every week that you'd hand to a robot tomorrow if you could? 3. How much time do you actually have this weekend? (a) about 4 hours total, (b) a full Saturday and Sunday, (c) one afternoon and that's it 4. Which AI are you using while we do this? (ChatGPT, Claude, Gemini, or something else.) Then save a file (or keep it at the top of this conversation if you can't make files) called WEEKEND-NOTES.md with my answers at the top. If I have less than 4 hours, tell me straight: do resources 1, 4, and 5 only, and skip the rest until next weekend. ━━━ STEP 2: BEFORE EACH RESOURCE (1 minute) ━━━ When I tell you I'm starting one, give me exactly three things and nothing else: - One sentence on why this one matters for the job I named in question 2 - Two things to watch for while I go through it, written as questions I should be able to answer afterward - A guess at how long it'll take me Then get out of the way and let me go do it. ━━━ STEP 3: AFTER EACH RESOURCE (5 minutes, the part that matters) ━━━ When I come back, run this, one question at a time: 1. "In your own words, what was that about?" If I give you a vague answer, push once. "It was about agents" is not an answer. 2. Ask me the two questions you gave me up front. If I can't answer one, tell me the timestamp or lesson to go back to instead of just telling me the answer. 3. Ask me ONE question that forces me to apply it to my real work. Not trivia. Something like: "Based on that, which part of [the task I named] would you hand to an agent first, and which part would you keep?" 4. Write the single most useful thing I said into WEEKEND-NOTES.md under that resource's name, in my words, not yours. 5. Tell me what percent of the weekend I've finished, and what's next. If I got something wrong, correct me directly. Don't be nice about it. Being wrong on Saturday is free. Being wrong on Monday in front of my boss or my client is not. ━━━ STEP 4: THE RUNNING LIST ━━━ Every time I say something like "oh, I could use that for X," stop and add it to a list at the bottom of WEEKEND-NOTES.md called BUILD IDEAS. Add the idea in my own words plus one line on what it would save me. Don't discuss it, just capture it and keep going. We use this list on Sunday night. ━━━ STEP 5: END OF EACH DAY (2 minutes) ━━━ Saturday night, ask me these and save the answers: - What clicked today? - What's still fuzzy? - What's the one thing you'd tell a friend who knows nothing about AI? Then tell me exactly what I'm starting with tomorrow, and stop. No pep talk. ━━━ WHEN I'M DONE WITH ALL 5 ━━━ Say this and nothing more: "You're done with the resources. Open the Sunday night build prompt and paste it in a NEW conversation, along with your WEEKEND-NOTES.md file."
The 5 Resources
In this order, for a reason
Every one of these is free, and not one of them is mine. Links go straight to the source.
Saturday morning
60 min video, 90 min with notes
Building AI Agents that actually work
Greg Isenberg and Remy Gaskell
Start here because it gives you the mental model everything else hangs on. An agent is just AI that can take actions instead of only talking. Most people never get this straight, so they spend months typing at a chatbot and wondering why nothing in their life changed.
Do this: Watch it with a notepad open. Every time he describes something an agent does, write down the closest version of that in your own work. You'll have 5 or 6 by the end and one of them is what you build Sunday night.
You walk away with: The difference between a chatbot and an agent, and which one you actually need.
Saturday afternoon
2 hrs
AI Fluency: Framework and Foundations
Anthropic, free course
This is the one people skip, and it's the one that separates the crowd. It teaches the 4D framework, which is a way of working with AI rather than prompting it and hoping. Free, straight from the company that builds Claude.
Do this: Do it inside your own AI account, not in a second tab you're half reading. Every time the course gives you an exercise, run it on a real task from your actual week. Fake practice tasks teach you almost nothing.
You walk away with: How to delegate to AI the way you would to a person, instead of gambling with prompts.
Sunday morning
60 to 90 min
Microsoft's AI For Beginners
Microsoft, 24 free lessons on GitHub
24 lessons on how AI actually works under the hood. You do not need to code to read it, and you should not try to finish all 24 this weekend. This is the resource that stops you being fooled by hype, because you'll finally know what the machine is and is not doing.
Do this: Read the intro section, then pick 3 lessons on whatever you keep hearing about and not understanding. Ask your AI to explain any lesson you bounce off of, then go back and read it again. Bookmark the rest for later.
You walk away with: Enough of what is under the hood that you can tell a real claim from a sales pitch.
Sunday morning
45 to 60 min
Prompt engineering interactive tutorial
Anthropic, learn by doing
Prompting is the one skill that transfers to every AI model, now and later. Learn it once and it keeps paying. This tutorial makes you write the prompts instead of watching someone else write them, which is why it sticks.
Do this: Not technical? Skip the setup entirely. Read each lesson right on GitHub and try the examples in a normal chat window. The lessons are the point, the code is not.
You walk away with: The handful of moves that turn a mediocre answer into a usable one, on any model.
Sunday afternoon
60 min
Become AI Native in less than 60 mins
Greg Isenberg and Theo Taba, Startup Ideas podcast
Finish with this one, not start with it. Now that you know what agents are and how to prompt, this is the episode that ties it into how people and companies actually reorganize their work around AI. If you feel behind, this is the hour that fixes it.
Do this: Watch for the workflows they walk through, not the big picture talk. Pause on each one and ask yourself which piece of your week it maps to. Those answers go straight into tonight's build.
You walk away with: What working AI native actually looks like day to day, beyond the buzzword.
Sunday Night
Build one thing before you go to bed
This is the part that decides whether the weekend mattered. Everyone who learns AI and stays stuck has the same thing in common: they consumed and never built.
Two hours. One small thing. Finished and used once tonight, not a plan for something bigger later.
Open a new conversation, paste this prompt, and give it your WEEKEND-NOTES.md file if you made one. It pulls three candidates from the ideas you collected, tells you which one to build and why the others lost, cuts it down to something you can actually finish, then walks you through it one click at a time.
Copy-paste prompt 2 of 2
SUNDAY NIGHT BUILD I just spent a weekend learning AI from 5 free resources. Tonight I want to prove I actually learned something by building one small thing that works, in about 2 hours, and using it once before I go to bed. If I have a WEEKEND-NOTES.md file, read it first, especially the BUILD IDEAS list at the bottom. If I don't, ask me the 3 intake questions below instead. Assume I'm smart but brand new. I have never written code. Explain any technical word the first time you use it, in one plain sentence. ━━━ RULES ━━━ - Ask ONE question at a time and wait for my answer. - Start every message with a progress tag like [Pick 2 of 3] or [Build: step 4 of 6]. - Watch the clock. Tell me the time estimate before each step. If we hit 90 minutes and it isn't working yet, stop adding and cut the plan down to whatever does work. - Small and finished beats big and broken. Every single time. - Never invent facts, numbers, prices, or quotes about me or my work. Ask me or use a connected tool, and say where it came from. - Don't write me a plan for something I'd build "later." Tonight only. ━━━ STEP 1: INTAKE (skip anything WEEKEND-NOTES.md already answers) ━━━ One at a time: 1. What do you do for work, school, or your business? 2. What's a task you did more than once in the last 2 weeks that felt like a copy of the time before? 3. Where are we working? (a) Claude Cowork or Claude Code, where you can make real files on my computer, (b) a regular chat window, (c) not sure If (c), ask me what app I'm in and figure it out from there. If we're in a regular chat, say so plainly: we can still build something real tonight, it just has to be something that lives in the chat or in a document rather than a file on my machine. ━━━ STEP 2: PICK THE ONE THING (10 minutes) ━━━ Pull 3 candidates from my BUILD IDEAS list and my answers. For each one, tell me in 2 lines: - What it would do - What it gives me back per week, in minutes or dollars Then score each one out of 5 on: can we finish it tonight, will I actually use it on Monday, does it use something I learned this weekend. Recommend ONE and say why the other two lost. I get the final call, but give me a real opinion, not a menu. Hard rule on scope: version 1 has to be something I could finish in 2 hours and use tomorrow. If my pick is too big, don't refuse it. Carve off the smallest slice that still does one useful job, build that tonight, and write the rest down for next weekend. Good size for tonight: - A skill or saved prompt that does a task I repeat, my way, every time - A one-page personal site that says what I do - A small tool just for me, like a tracker or a checklist that remembers - An agent that takes one input and gives one output, like turning meeting notes into a follow-up email in my voice - A document or template I'll reuse weekly, built properly Too big for tonight: anything with user logins, payments, a database of other people's data, or the word "platform." ━━━ STEP 3: THE ONE-PARAGRAPH BRIEF (5 minutes) ━━━ Before you build anything, write this back to me and get a yes: - What it does, in one sentence a 10 year old would understand - What is NOT in version 1 (list at least 3 things we're cutting) - The test that proves it works, written as something I do, not something you claim. Example: "Zach pastes last Tuesday's meeting notes in and gets back an email he would actually send." Save it as BUILD-BRIEF.md if you can make files. If I say no or change something, rewrite it and ask again. Do not start building without a yes. ━━━ STEP 4: BUILD IT (the rest of the time) ━━━ Now build, in steps. For each step: 1. Tell me what you're about to do and why, in 2 lines. 2. Do it. 3. Tell me exactly what to click, type, or paste, one action at a time. Never assume I know where a button is. 4. Show me the result and ask if it looks right before moving on. Use what I learned this weekend, and name it when you do. If we're using a prompting technique from the Anthropic tutorial, say so. If we're structuring this the way the agents video described, say so. I want to see the connection between what I watched and what we're making. If something breaks: ask me for the exact error message or a screenshot before you touch anything. Then explain what went wrong in plain English, fix one thing, and tell me how to check that it worked. Never fix three things at once. ━━━ STEP 5: USE IT ONCE, TONIGHT (10 minutes) ━━━ Not a demo. The real thing, with my real input. Run the test from the brief. If it fails, we fix it now. If it passes, tell me: - Exactly how to open or run it again tomorrow, written so I could follow it half asleep - The one thing most likely to break it, and what to do when it does ━━━ STEP 6: CLOSE OUT (5 minutes) ━━━ Write NEXT-BUILDS.md with the 2 ideas we didn't pick and the pieces we cut from tonight's build, so I have a starting point next weekend. Then give me a final message with exactly this and nothing else: - What I built, in one line - What it gives me back per week - The one sentence I'd use to explain it to someone who asks what I did this weekend
Short On Time
The two hour version
Only have one afternoon? Do these three things and nothing else. You will get most of the way there.
Watch resource 5 first, not resource 1 (60 min)
If you only get one hour of input, make it the catch up hour. It covers the most ground per minute.
Do the first third of the prompt tutorial (20 min)
The early lessons carry most of the value. You can feel the difference in your own chats the same day.
Run the Sunday night build prompt anyway (40 min)
Tell it you only have 40 minutes when it asks. It will shrink the build to fit. Something small and finished still counts.
What Goes Wrong
5 things that quietly ruin this weekend
Watching all five back to back
You'll remember almost none of it. The 5 minutes of quizzing after each one is where the learning actually happens, which is exactly what the study partner prompt forces you to do.
Taking notes you never open again
Notes in a notebook are dead weight by Monday. Notes in a file your AI can read become the raw material for Sunday night. That is the whole reason WEEKEND-NOTES.md exists.
Trying to finish all 24 Microsoft lessons
That is a multi week course, not a Sunday morning. Three lessons you understand beats twenty four you skimmed. Bookmark the rest without guilt.
Skipping the prompt tutorial because it lives on GitHub
GitHub looks like a place for developers and it scares people off. You are reading lessons on a web page. That is all. No install, no terminal, no code.
Ending the weekend without building anything
This is the big one. Five resources and zero builds means you learned about AI instead of learning AI. Two hours Sunday night with one small finished thing beats another video every time.
Credit Where It Belongs
Who made these
None of these five resources are mine. The schedule, the study partner prompt, and the Sunday night build prompt are what I added. Everything else belongs to the people who made it and gave it away for free:
Greg Isenberg and Remy Gaskell for the agents course. Anthropic for both the AI Fluency course and the prompt engineering tutorial. Microsoft for AI For Beginners. Greg Isenberg and Theo Taba on the Startup Ideas podcast for the AI native episode.
If a resource helps you, go say so wherever they posted it. That is how free stuff keeps getting made.
Then on Monday, use the thing you built. Not once to prove it works. Every time the task comes up.
That is the whole difference between people who spent a weekend learning about AI and people who came out of it with something running. Five resources, zero dollars, one weekend.
Want the longer version with a day by day pace instead? Learn 99% of Claude in 7 Days covers similar ground across a full week, with a bigger build at the end.
Work with Me
Want me on your problem for an hour?
Book a 1:1 call and we build the fix live — the workflow you want automated, the tool you can’t crack, whatever’s stuck. Direct, hands-on, no pitch waiting at the end.