Nobody Assigned This

Objectives

By the end of this session you can:

• Explain the difference between agency and intelligence in your own words, with one example from your own life.
• Name which low-agency belief shows up most for you, and in what situations.
• Record a baseline score on the 10 behaviors in your journal.

Hook: The Noun Hunt (3 min)

Look around the room you're in. Pick five objects: the chair, the charger, the light switch, the app you last opened, the paint color on the wall. For each one, say out loud: "A person decided this should exist."

Now pick one and guess: what did that person have to figure out that nobody had taught them?

Hold that thought. Someone famous said almost exactly this, and you're about to read it.

Reading: Agency > Intelligence (7 min)

In February 2025, Andrej Karpathy, a co-founder of OpenAI who used to run AI at Tesla, posted something surprising coming from a person known mostly for being brilliant:

"Agency > Intelligence. I had this intuitively wrong for decades… Agency is significantly more powerful and significantly more scarce. Are you hiring for agency? Are we educating for agency? Are you acting as if you had 10X agency?"

Are we educating for agency? Mostly, no. School rewards a different skill. Someone hands you a clear task, a rubric, and a due date, and you execute. You're good at that. It's a real skill. It's also the skill AI got cheap at first.

Agency is what happens when there's no assignment. Something is broken, missing, or could be much better, and you decide it's yours to act on. You don't wait for a teacher, a tutorial, or someone's blessing. You find a way.

Mitchell Hashimoto, co-founder of the software company HashiCorp, said it in September 2026:

"I can't stress enough how little an idea matters compared to the agency of the people executing the idea… The difference between mediocre and excellent work and outcomes is predominantly one of agency… They don't wait for someone else to do something, for someone to teach them, for someone to give them the path… They just go out and find a way to do it… Look around the room you're sitting in… Every noun. It almost all exists because a person willed it into existence. Nothing is stopping you from doing the same."

George Mack, who wrote an essay called High Agency, boils it down to two triangles of beliefs:

Look closely at the low triangle. None of it is about being dumb or lazy. It's all uncertainty and fear. Smart people live there constantly, sometimes more than anyone, because they're used to knowing the answer before they start.

Two things to clear up now. Agency has nothing to do with being loud, grinding until 2 a.m., or acting like the main character. Some of the highest-agency people you'll ever meet are quiet. And real agency includes judgment: knowing when to ask first, and owning it when you get it wrong. You'll practice both.

Agency is a behavior, which means it's trainable. You'll train it the only way that works: by doing things nobody asked you to do and seeing what happens.

Discussion (8 min)

Pick three.

1. Think of someone you actually know (not a celebrity) who has high agency. What did they do that made you notice?
2. Karpathy says agency is "more scarce" than intelligence. What in your normal week trains intelligence? What, if anything, trains agency?
3. Which low-agency belief is loudest for you? Is it the same at school, with friends, and online?
4. Pushback: "Nothing is stopping you" sounds great, but plenty of people lack money, time, connections, or safety. Is Hashimoto's line true? Write the most honest version of it you'd actually believe.
5. When is "I will figure it out" the wrong attitude?

Exercise: Baseline + Stuck List (10 min)

1. Baseline (4 min). Open your Reflection Journal and score yourself 1–5 on all 10 behaviors in the "Before" column. Be honest; nobody grades this.

2. Stuck List (6 min). Write three things you've wanted to do, make, or change for more than a month and haven't. Next to each, tag the belief that's holding it:

   • D = "I don't know what to do"
   • P = "I'm afraid of what people will think"
   • W = "I'm afraid it'll go wrong"

   Circle one. Write the smallest first step that would take under 15 minutes. You don't have to do it yet. Seeing how small it is usually does something.

Challenge: The Unasked Fix (30–60 min, before Session 2)

Find something at home that's been broken, annoying, or undone for at least a month, that nobody has asked you to fix. Fix it.

Ideas: the drawer that won't close, the remote that needs batteries, the family calendar chaos, the photo backup nobody set up, the Wi-Fi that dies in one room, the junk drawer, the bike with the flat tire.

Rules:

• Don't announce it first. Don't ask anyone how. Figure it out (search, video, AI, trial and error).
• Once it's done, tell the person it affects in one sentence.

Done when:

• The thing is fixed and actually works.
• Three lines in your journal: what it was, what you had to figure out, what you were thinking right before you started.
• Problem Log started. For the next two days, write down every problem you notice at home, school, your team, your town, or online. Aim for 15. Tiny counts. Complaints count. This list feeds your capstone.

Stretch: Fix something outside your house: the team equipment bag, a broken link on your club's page, the lost-and-found. If it isn't yours, get the owner's OK first.

Safety: Nothing electrical beyond batteries and plugging things in. No tools you haven't used before without an adult nearby. No spending money without asking.

Journal

• What were you telling yourself in the 10 seconds before you started the fix?
• Which point of the high-agency triangle did you use? Which one was missing?
• How much did you actually need help, compared with how much you thought you would?

Objectives

• Classify any message about a problem as Level 1 through Level 5.
• Rewrite a Level 1 complaint into a complete Level 4 message.
• Decide when Level 5 is appropriate and when Level 4 is the ceiling.

Hook: Three Texts (3 min)

Three texts to a parent. Rank them best to worst.

A: printer isn't working

B: printer's out of black ink. the spare in the hall closet is the wrong model. right one is $28 at the office store today or $22 online tomorrow. my lab report prints Thursday so tomorrow works. want me to order it on your account?

C: printer was out of ink, put in the spare, printed my report. we're out of spares now so I added it to the shopping list.

Which one made the parent do the work? Which one would you want to receive?

Reading: The 5 Levels of Work (6 min)

Alex Lieberman, co-founder of Morning Brew, shared a framework he credits to Steph Smith. Every time you bring someone a problem, you're operating at one of five levels:

Lieberman tells new hires to live at Level 4 from day one and earn Level 5 through trust.

That's the key distinction. Level 4 is where you do all the thinking, but the decision belongs to someone else because it's their money, their stuff, or their call. Level 5 is for things inside your own authority, or where someone has told you "just handle it." Text B above is Level 4. Text C is Level 5, and it works because a printer cartridge the family already owns is fair game.

Three common mistakes:

• Level 1 with extra feelings. "The printer is ALWAYS broken" is still Level 1.
• Fake Level 4. Listing options with no pick: "we could do X or Y, idk, what do you think?" Picking is the hard part. Do the hard part.
• Level 5 overreach. Fixing something that wasn't yours to fix, like reorganizing your sister's room. If the other person would reasonably want a say, it's Level 4.

Agency starts one step before any of this. Most people notice problems and file them as weather: "that's just how it is." The high-agency move is to ask whose problem is this? The honest answer is often "partly mine, if I want it to be."

Discussion (7 min)

1. What level do you usually operate at with parents? Teachers? Friends? Why the difference?
2. When is Level 1 exactly right? (Think: danger, someone getting hurt, an emergency. Say it fast.)
3. Pushback: At school, isn't Level 5 sometimes just doing the teacher's job, or being a know-it-all? Where's the line?
4. What problem does everyone around you treat as weather?

Exercise: Level-Up Drill (12 min)

Rewrite each complaint as a Level 4 message using this template:

Problem (one sentence, with a fact) · Cause (what you think and how you know) · Options (2–3) · My pick + why · What I need from you · By when

1. "The group chat for our project is chaos."
2. "There's never anything for lunch."
3. "Robotics never gets enough practice time."
4. One from your own Problem Log.

Worked example for #2:

"Lunch keeps falling apart on school days. I skipped it twice this week because we were out of anything I could pack in five minutes. I think it's because nobody plans lunches when the groceries get ordered. Options: (a) I plan my own lunches and add them to the list every Sunday, (b) we buy a few default items in bulk, (c) I buy lunch at school two days a week. I'd pick (a). It costs nothing extra and it's on me. I need you to let me add to the order before it goes in. Can we start this Sunday?"

Then circle any problem on your list you could take to Level 5 without anyone's permission. That's a candidate for your challenge.

Challenge: Level Up (45–60 min, before Session 3)

Part A: Level 4, delivered. Pick a real problem from your Problem Log that sits in someone else's authority: a parent, teacher, coach, club leader, or manager. Spend at least 20 minutes investigating (count things, ask two people, look it up). Deliver your Level 4 in person, by text, or by email to the person who can actually say yes.

Part B: Level 5, done. Pick a small problem entirely inside your authority. Fix it. Then send the one-line "keeping you in the loop" message to whoever it affects.

Example Level 5 message: "Heads up: the team ball pump needle was broken, so I bought a replacement set ($4, my money) and put it in the bag. Nothing you need to do."

Done when:

• Level 4 delivered, and the response is logged (yes, no, and "let me think" all count).
• Level 5 done, and the loop-in message sent.

Stretch: Your Level 4 goes to an adult outside your family.

Safety: For teachers or coaches, use school email or official channels.

Journal

• How did the person react to your Level 4? Were they surprised? What does that tell you?
• What was the hardest part: finding the cause, picking an option, or sending it?
• Where were you tempted to stop at Level 2?

Objectives

• Name the specific people whose judgment you fear for one real goal, and notice what changes when you do.
• Sort situations into "ask forgiveness" or "ask permission," and defend each call.
• Draft a cold email a busy adult could say yes to in under two minutes.

Hook: Who Is "They"? (3 min)

Think of something you haven't done because of what people would think. Write it down.

Now write the actual names of the people you're worried about. Not "people." Not "everyone at school." Names.

Count them. George Mack observes that it's usually only 1 to 7 specific people, and naming them cuts the fear roughly in half. Check yourself: did it?

Reading: The Permission Slip Doesn't Exist (7 min)

For decades, aerospace companies stopped building supersonic passenger jets. Blake Scholl, founder of Boom Supersonic, put it this way in July 2026:

"For 50 years, no one in aerospace was willing to build a supersonic jet. They were all waiting for regulations to change… It turns out that building a supersonic jet is how you change the regulations."

Reddit co-founder Alexis Ohanian: "The permission you're waiting for doesn't exist until someone builds it into existence." Jason Fried of 37signals describes entrepreneurship as "'Ok, that gives me an idea' and then you do it. You see what someone else would miss, you do what no one would give you permission to do." Shopify CEO Tobi Lütke says it in five words: "You can just build things."

DHH, who created Ruby on Rails, adds: "Agency is rarely given. It's taken, then recognized." Taken here means you act like it's yours to do before anyone confirms it, and the recognition comes afterward.

Sounds like a license to do anything. It isn't, and this is where judgment comes in. Erik Torenberg of a16z offers a useful test:

• Ask forgiveness when the upside is big and the worst case is that you look a bit foolish (and it's ethical).
• Ask permission when the downside could hurt a relationship or your reputation.

Add a teenager's version: anything involving safety, money, other people's stuff, other people's private information, or speaking for a group needs permission first. Everything else, try it.

On asking. Most adults who've built something remember someone who answered their email when they were young. Plenty of them would happily pay that forward. They can't if you never ask. What gets a yes:

• Short. Under 150 words. Under 50 for a DM.
• Why them, specifically. Prove you know their work.
• A tiny, concrete ask. One question. Feedback on one thing. Fifteen minutes. Never "be my mentor."
• Easy to say no to. No guilt, no pressure, no "I'll follow up daily."
• A clear subject line that says what it is.

Bad:

Subject: Hi! Hi, my name is Maya and I'm 15. I'm really passionate about technology and I want to learn everything I can. I really admire you. Could you be my mentor? I'd love to pick your brain sometime. Let me know! Thanks so much!!!

It could be sent to anyone. "Mentor" is an open-ended commitment. "Pick your brain" means do unpaid work for me. No easy yes exists.

Good:

Subject: One question from a 15-year-old about your bike-lane project

Hi Ms. Alvarez,

I'm a high school sophomore in [town]. I read your piece in [local paper] about the Elm Street bike lane, especially how you got the city to run a 30-day trial first. I'm trying to get my school to add bike racks near the gym. Right now 40+ bikes lock to a fence.

One question: when you pitched the trial, what made the city say yes? Even two sentences would help a lot.

Thanks for reading, [First name]

DM version (under 50 words):

Hi [Name], I'm 15 and your video on [specific thing] is why I started [specific thing]. One question: [specific question]? Totally fine if you're too busy. Thanks for making it.

Template:

Subject: One question from a [age]-year-old about [their specific work] Hi [Name], I'm a [grade] in [town/state]. I [read/watched/used] your [specific thing], especially [specific detail]. I'm working on [your project, one sentence, with one real fact]. One question: [single answerable question]? Even a two-line answer would help. Thanks, [First name]

Discussion (6 min)

1. Who's on your "they" list? What would each of them actually do if you tried the thing and it flopped?
2. DHH says agency is "taken, then recognized." When is taking it fine, and when is it just rude?
3. Pushback: Scholl's story works because he built a working company. Isn't "ignore the rules and build it" dangerous advice for a 15-year-old? Where exactly does the forgiveness/permission line sit for you?
4. Who's someone you'd email if you knew they'd read it?

Exercise (12 min)

Part 1: Forgiveness or permission? (4 min) Decide each one fast, then compare with your mentor.

1. Fixing typos on your club's public website, where you already have edit access.
2. Posting a video of your friend's wipeout for views.
3. Emailing an author you admire with one question.
4. Rearranging the living room while your parents are out, to make a study corner.
5. Starting a lunchtime chess meetup in the school library.
6. Buying a $15 domain for your project with a parent's card.
7. Signing up to speak during public comment at a town council meeting about a dangerous crosswalk.
8. Answering a local reporter's question on behalf of your club.

(Discuss with your mentor. The suggested answers are in the parent's answer key.)

Part 2: Draft your ask (8 min). Pick a real person for your challenge: someone within reach, whose work you actually know. Good targets: a local business owner, a researcher at a nearby university, a writer, an engineer, a college student doing what you want to do, a creator with under 50,000 followers. Draft the email using the template.

Challenge: The Ask (45–75 min, before Session 4)

Part A. Send one cold email or DM to someone you admire, after a parent reviews it.

Part B. Make one in-person ask that makes you slightly uncomfortable. Examples: ask a shop owner if they'd post a flyer for your project, ask a teacher whether you can do an assignment in a different format, ask your coach for one specific thing to fix, ask a store if they have a student discount.

Done when:

• Part A is sent, and a copy is in your Ask Log. A reply isn't required. Sending is the skill.
• Part B is done, and the outcome is in your Ask Log.
• Quiz 1 done (10 min, no notes, no AI).

Stretch: Send three cold emails to three different people on the same day. Track the reply rate.

Safety:

• A parent reviews every message to someone you don't know before it goes out.
• Use a parent-approved email or account. Be honest about your age.
• Never share your home address, school schedule, phone number, or photos of yourself.
• Don't agree to a call or meeting unless a parent arranges it and is present.
• If any reply feels off, stop replying and show a parent.

Journal

• What happened to the fear when you wrote down the actual names?
• What was your most ridiculous-in-hindsight worry before you hit send?
• Did you make a forgiveness/permission call this week you'd make differently now?

Objectives

• Explain why AI makes agency more valuable, using at least two ideas from the reading.
• Direct an AI through a task that would normally take a full day, then judge and improve its output against criteria you wrote first.
• Pass the "I did this" test by naming three decisions that were yours.

Hook: Dead Code (3 min)

Read this from Paul Graham, co-founder of Y Combinator (June 2026):

"College students use AI to do most of their writing. An increasing number of professors secretly use it for grading. In the limit case, AIs do all the work, and all the humans do is transmit what they create. A good compiler would recognize this as dead code and remove it."

A compiler turns code into a program and deletes lines that don't do anything. So in Graham's scenario, who is the dead code?

Take 60 seconds. Then answer honestly: where in your own life are you, right now, the dead code?

Reading: When Doing Gets Cheap (6 min)

Dickie Bush, founder of Ship 30 for 30: "In a world of abundant intelligence where anyone can build anything, initiative becomes the most valuable skill." His description of how high-agency people use AI: they "decide what to do themselves, then use AI to help them achieve it." Decide first. Then build.

Sahil Bloom, author of The 5 Types of Wealth, points out that the old career recipe (predict which skills will be valuable, learn them, cash them in) works less well now:

"This 'new world' will reward those who are just willing to be beginners over and over again. Value now seems to accrue to those who can pair insatiable curiosity with bias for action. The valuable skills of the future aren't really fixed skills. They're much more akin to mindsets."

Wharton professor Ethan Mollick studies how people use AI, and he's noticed that most don't aim high: "people don't try ambitious things with these systems. Many models are excellent as a Google replacement, for homework 'help,' etc. It is someone's agentic use of frontier AI on long-horizon real problems that is impactful." Homework help is the floor.

Dan Shipper, CEO of Every, describes a feeling you may have had: when AI suddenly does a task you used to do, it can feel like an "agency rupture," as if the thing that made you useful just evaporated. The people who thrive rebuild. AI "fast-forwards you through the boring or repetitive parts of the work," and the direction and quality control become "the actual work itself—the interesting part." You go from saying "The AI did this" to "I did this."

Then there's Shaan Puri's theory (paraphrased). People who were teenagers around 1998–2004 accidentally became very high-agency because nothing was handed to them. Getting music, finding information, and learning skills meant digging through obscure forums and figuring things out alone. Today almost everything arrives on demand, including answers. So you have to build agency on purpose. This course is you doing that.

Direct vs. delegate

One warning: AI can be confidently wrong. Anything that matters (a fact, a price, a date, a claim about a real person) gets checked against a real source.

Discussion (5 min)

1. Where are you currently a pipe for AI? Where are you the director?
2. Bush lists things that get more valuable with "infinite intelligence": courage, agency, community, accountability, judgement, attention spans, personality. Which do you have the most of? The least?
3. Pushback: If AI does the work and you only direct it, did you learn anything? When should you do something the slow way on purpose?
4. What's the most ambitious thing you could try with AI in the next 10 days?

Exercise: The Day-to-Hour Sprint (18 min)

Pick a task that would take you a full day alone:

• A one-page website for something real (your club, your capstone idea, a neighbor's small business).
• A comparison of four options for a decision your family actually faces (summer programs, phone plans, a weekend trip), ending with a recommendation.
• A study guide plus a 20-question practice test for your next unit (for studying, not submitting).
• A full event plan: budget, schedule, checklist, invite message.

1. Decide, no AI (4 min). Write down: who it's for, three criteria for "great," and one thing it must not do.

2. Direct (6 min). Write a real prompt. Include context, audience, constraints, and format. Ask the AI to ask you questions first.

Weak prompt: "make a website for my dog walking business"

Strong prompt: "I'm 15 and starting a weekend dog-walking service in my neighborhood. The audience is busy parents and older neighbors who'll scan a QR code on a flyer. Build a single-page HTML site with a headline, three services with prices ($12 for a 30-minute walk, $20 for 60 minutes, $8 for a feeding visit), a three-step 'how it works,' and a contact section that uses a parent-approved email only, no phone or address. Tone: friendly and trustworthy, not cutesy. Before writing any code, ask me up to five questions that would make this better."

3. Judge (5 min). Score the output against your three criteria. Mark what's wrong, generic, false, boring, or missing. Then try a critique prompt:

"Act as a skeptical [neighbor / principal / customer] seeing this for the first time. List the five things that would make you not trust it or not act on it. Be blunt."

4. Improve (3 min). One round of specific direction, or edit it yourself:

"Keep the headline and the pricing section. Cut the testimonials; they're fake and I won't use fake reviews. The 'how it works' section sounds like an ad. Rewrite only that section so it sounds like a real 15-year-old wrote it."

"I did this" statement. Write three decisions you made that the AI didn't.

Challenge: The Long Push (60–90 min, before Session 5)

Read the Capstone section. Pick your project. Fill in Capstone Brief v1. Then spend 60 focused minutes using AI on the real problem: research, a prototype, a plan, a first draft, a list of people to reach.

Done when:

• Capstone Brief v1 is filled in.
• One tangible thing exists: a prototype, a draft, a plan, or a list of 10 potential users and how you'll reach each one.
• An AI Ledger entry: what AI did, what you decided, and what you rejected.

Stretch: Put your brief in front of one adult outside your family for a five-minute reaction.

Honesty and safety:

• Follow your school's AI rules. Never submit AI work as your own for a grade.
• Don't paste other people's private information (names with details, their messages, their photos) into AI tools.
• Your parent knows which AI tools you use.

Journal

• What did the AI get wrong that you caught? What might you have missed?
• Which decision in the sprint was hardest? That's probably where the real work was.
• Finish the sentence: "I did this because I decided ___."

Objectives

• Cut your capstone down to a v0.1 you can put in front of a real person within 48 hours.
• Classify a "no" as not now, not this, or not me, and write the right follow-up.
• List what's likely to break and how you'd fix it.

Hook: The Tally (3 min)

Count every ask you've made since Session 1: fixes, Level 4s, cold emails, in-person asks. How many yeses? Nos? Silences?

If you got zero nos, you didn't ask for enough.

Now the real question: what did each "no" actually cost you? Minutes? A little pride? Anything real?

Reading: Small Bets (5 min)

The third point of George Mack's high-agency triangle is "I can fix it if it breaks." It's the belief that makes starting possible. You don't need a perfect plan if you trust yourself to repair what goes wrong. One of Dickie Bush's traits of high agency says the same thing from another angle: default to action over rumination.

The tech commentator signüll describes high agency as "some unholy combination of: resourcefulness, relentlessness, resilience," and adds: "the world has never had more leverage just sitting around waiting for one person stubborn enough to use it." DHH: "Every audacious mission looks impossible until it's done."

In practice, all of that shrinks down to one move: ship a v0.1. The smallest version a real person can react to.

The version you're slightly embarrassed by is the one that teaches you something. Nobody learns from a draft in a folder.

The three kinds of no:

• Not now. Timing. Ask "Would it be OK if I checked back in [month]?" and put it on your calendar.
• Not this. The offer is wrong. Ask "What would make it useful for you?"
• Not me. Wrong person. Ask "Is there someone you'd suggest I ask?" This is the best no to get.

Silence usually means a busy inbox. One polite follow-up after 5–7 days is normal. After that, move on.

Follow-up template: "Hi [Name], bumping this in case it got buried. Still hoping to hear [the one question]. If now's not a good time, no worries at all. [First name]"

Discussion (4 min)

1. What's the scariest part of showing someone v0.1?
2. Pushback: Some things shouldn't ship ugly: a bridge, medicine, an apology. Where's the line between "ship it ugly" and "sloppy"? Which side is your capstone on?
3. Relentless or annoying? How many follow-ups is too many, and does it depend on who you're asking?

Exercise (11 min)

1. The Cut (5 min). List everything your capstone "should" have. Cross things off until what's left could be shown to a real person in 48 hours.
2. Break List (3 min). Three things likely to break or go wrong, with one fix for each.
3. No-Sorter (3 min). Look at every no or silence in your Ask Log. Label each one (not now / not this / not me / silence) and draft the follow-up for one of them.

Challenge: Ship v0.1 (60–90 min, before Session 6)

Put v0.1 in front of at least one real person outside your family. Watch them use it, or walk them through it. Then ask three questions and write down the answers word for word:

1. What was confusing?
2. What would make you use this, come back, or tell someone about it?
3. What did you expect to see that wasn't there?

Also: if it's been five or more days since your cold email with no reply, send one follow-up.

Done when:

• A real person outside your family has used or seen v0.1.
• Their feedback is in your Ship Log, word for word.
• You've decided on one change.

Stretch: Three people. Or get one to commit: sign up, attend, share it, or pay.

Safety: Test with people you know through school, family, or your community, in public places or with a parent aware of the plan. No collecting money unless a parent sets up how.

Journal

• What did the person do that you didn't expect?
• When did you want to explain or defend instead of listening?
• What could you cut even further?

Objectives

• Rewrite a blame statement as an ownership statement without taking blame that isn't yours.
• Repair a mistake using the four-part repair.
• Agree with a parent on a real "use your judgment" zone.

Hook: Three Excuses (3 min)

"My partner didn't do their part." "The teacher never explained it." "The site crashed right before the deadline."

Each one might be completely true. For each, find the part that was still yours. (Maybe you knew on Tuesday that your partner was checked out. Maybe you never asked a question. Maybe you planned to submit ten minutes before the deadline.)

Reading: Ownership, and When Agency Goes Wrong (6 min)

Tim Ferriss, author and investor, in August 2026: "It's amazing how someone's IQ seems to double as soon as you give them responsibility and indicate that you trust them." He told his support reps that if a problem cost under $100 to fix, "use your judgment and fix it yourself… Don't ask me for permission." The emails he had to deal with dropped from 200 a day to under 20 a week. Nobody got smarter. They got trusted, and they rose to it.

Dickie Bush lists two habits of high-agency people that are hard to practice: never blame, complain, or defend, and treat outcomes as your responsibility.

Ownership doesn't mean everything is your fault. It means asking one question after anything goes wrong: "Given what happened, what's mine to do now?" Sometimes the answer is small. It's never "nothing."

Ownership also means keeping promises, including small ones. Every kept promise is a deposit of trust, and trust is what earns you Level 5.

When you get it wrong, use the four-part repair:

1. What I did. No "mistakes were made." "I did X."
2. The impact. "That meant you had to Y."
3. The fix. "I've already Z" or "Here's how I'll fix it."
4. Prevention. "Next time I'll W."

"I said I'd send the slides Tuesday and sent them Thursday. That left you one night to practice your part. I've already added speaker notes for your section. Next time I'll send a draft a day early, even if it's rough."

Now the fair critique. "High agency" gets misused, and you should recognize the knockoffs:

• Hustle theater. Lots of visible busyness, very little actually solved.
• Main character mode. Treating other people as extras in your story. Taking over the group project. Never asking whether anyone wanted your fix.
• Recklessness. Ignoring real downsides for other people: safety, money, privacy, reputation.
• Arrogance. "I'm high agency, so rules are for other people."

Real agency makes other people's lives easier, and it includes saying "I got that wrong."

Discussion (5 min)

1. Ferriss says IQ "seems to double" with responsibility. Have you seen that happen in yourself or someone else?
2. Where's the line between owning outcomes and blaming yourself for everything?
3. Pushback: Some things really are someone else's fault, or unfair by design. Does "never blame or complain" ignore that? How do you hold both at once?
4. Which knockoff version of agency are you most at risk of? Be honest.

Exercise (9 min)

1. Flip three (4 min). Take three real complaints from your week. Rewrite each as "what's mine to do."
2. Commitment ledger (2 min). List every promise you made this past week, including to yourself. Mark each kept, broken, or pending. For any you're going to break, renegotiate now, before the deadline.
3. Draft your zone (3 min). Write a "use your judgment" zone to propose to your parent:

Domain: Dinner on Wednesdays Limit: $40 in groceries, and a meal the whole family will eat I decide: What we eat, what to buy, when to cook Report back: A text with what I spent, by Wednesday night Off limits: Anything involving the stove unless an adult is home

Other domains: family tech support, a sibling's birthday plan, weekend trip research, pet care, your own school supplies budget.

Challenge: Run the Zone + v0.2 (45–75 min, before Session 7)

Part A. Agree your zone with a parent, in writing (a text counts). Make at least one real decision in it before Demo Day and log it in your journal.

Part B. Build v0.2 of your capstone based on your v0.1 feedback. Show it to at least one person (the same one or a new one).

Done when:

• Zone agreed in writing.
• At least one decision made inside it and logged.
• v0.2 shipped to at least one person, with their reaction logged.

Stretch: Own a mistake out loud this week using the four-part repair.

Journal

• What was the first decision you made in your zone? Did you want to ask permission anyway?
• Which promise did you almost break this week? What made the difference?
• What did v0.2 fix? What did it reveal that v0.1 hid?

Objectives

• Send a Level 4 or Level 5 status update on your capstone to your mentor.
• Name your biggest blocker and a way around it within 48 hours.
• Draft the outline of your five-minute demo.

Hook: One Sentence (2 min)

Say it out loud, in one sentence: what did you build, who is it for, and has anyone used it?

If that sentence was hard, that's useful information.

Status Update (5 min)

Write your mentor a real update. Level 4 or 5 only.

Level 5 status template: "Capstone update: [what exists now]. [Number] people outside my family have used it. The main feedback was [X], so I [changed Y]. Next: [one thing] by [day]. The one thing I need from you: [specific ask, or 'nothing']."

Unblock (8 min)

What's your biggest blocker? Work down this ladder until something moves:

1. Try a different route. If the front door is locked, find the side door. That's what relentless means in practice.
2. Ask someone who's done it. Dickie Bush's rule: learn from people ahead of you. A teacher, an older student, the person who answered your cold email.
3. Ask AI for ten workarounds. Then you pick one.
4. Shrink the scope. What's the version you can finish in 48 hours?
5. Move the deadline honestly. Tell your mentor now, Level 4 style, rather than on Demo Day.

Demo Draft (8 min)

Outline your five minutes using the demo format in the Capstone section. Bullets only, no script.

Challenge: Finish + Rehearse (60–90 min, before Session 8)

Done when:

• Your capstone is in its final shipped state for this course (it can keep going after).
• You've rehearsed the demo twice, once with a timer.
• You have one more real data point: a new user, a quote, a number, a decision someone made.
• Quiz 2 done (10 min, no notes, no AI).

Stretch: Invite your outside person (the user, customer, or advisor) to Demo Day, in person or by video, arranged through your parent. Or record a short thank-you and send it to them.

Journal

• What was your blocker, and which rung of the ladder got you past it?
• What's the part of your project you're proudest of? The part you'd hide if you could?

Objectives

• Present your capstone in five minutes using the demo format.
• Score yourself with the rubric and compare it with your mentor's score.
• Commit, in writing, to one thing you'll do in the next 30 days.

Setup (2 min)

Audience: your parent or mentor, and ideally your outside person, in person or on video (parent-arranged). If nobody else can come, record it.

Demo (5 min)

Use the five-part format from the Capstone section. Timer on. When it rings, stop.

Q&A (5 min)

Your audience asks. A few good questions for them to choose from:

• "What would you do differently if you started over tomorrow?"
• "What's one decision you made that the AI would've gotten wrong?"
• "What did the person outside your family actually say?"
• "Where did you want to quit?"

Scoring (6 min)

You and your mentor each score the rubric separately, then compare. Talk about the biggest gap.

After-Rating (5 min)

Fill in the "After" column on all 10 behaviors in your journal. Don't look at your "Before" scores until you're done.

Retro + Next 30 Days (7 min)

Answer out loud:

1. What worked?
2. What didn't?
3. What surprised you about yourself?

Then write one commitment for the next 30 days. Make it specific, doable, and at Level 4 or 5.

Examples: "Keep the practice tracker running for the team through the end of the season." "Send one cold email every Sunday." "Take the bike-rack proposal to the principal with the survey data."

Post it where you'll see it. Tell one person.

Journal

• Which behavior score moved the most? Why do you think?
• What would the version of you from Day 1 think of what you shipped?
• What's the next thing nobody's going to assign you?