← Exercises

start a startup today

deadline: next Saturday’s class · goal: earn at least $1 online · groups of 3–4

You are now living in the future. Because of the last three weeks, frontier AI tools are so new, and changing so quickly, that every problem, confusion, pain point, or delight you have experienced is something ~8 billion other people probably haven’t encountered yet.

This is like when the iPhone launched, when the personal computer revolution happened, when the internet popularized. New platforms create new opportunities for people to solve each other’s problems. Smartphones made Airbnb, Uber, Instagram, Snapchat, and DoorDash possible. The internet made Google, Yahoo, Amazon, eBay, Craigslist, Myspace, and Facebook possible.

AI is opening up a new wave of problems that can be solved — and you are experiencing those problems first, so your hunches about good ideas are probably right.

Paul Graham says the best startup ideas commonly satisfy three conditions:

  1. The founders themselves want it.
  2. The founders can build it.
  3. Few other people realize it is worth doing.
“If you’re at the leading edge of a field that’s changing fast, when you have a hunch that something is worth doing, you’re more likely to be right.”

So far, we’ve looked at your own problems and asked you to solve them with agents. Today, we’re going to teach you how to share those solutions with others — and in doing so, make your first dollar(s) on the internet. Maybe this becomes the start of a side hustle, or a startup.

Earn one dollar by next Saturday’s class to win the game. The person with the most net-new revenue earns a framed photo on the (to be created) Fractal hall of fame.

1. ideate

in groups, ask your agent, then share (10 min)

Ask the agent that knows you. Use your personal agent — its memory of your life and work is the advantage. List its best ideas in the worksheet.

Give Me 10 First-Dollar Ideas →
10:00

what’s frustrating? (5 min)

Each person lists three frustrations at the AI frontier, then discuss. Ask yourself:

  • Where does your agent misunderstand you?
  • What keeps breaking?
  • What just isn’t working?
  • What feels too technical?
  • Where do you distrust your agent?
5:00

what’s wonderful? (5 min)

Each person lists three wonders at the AI frontier, then discuss. Ask yourself:

  • What is the best experience you’ve ever had with your agent?
  • What do you love?
  • What is the most helpful prompt or conversation you have ever had?
  • What do you wish you could do, that seems possible and maybe wondrous?
5:00

choose the idea (3 min)

Write the one idea you’re excited to pursue this week in the worksheet.

3:00

2. structure the offer (10 min each)

Take turns interviewing each member of your group to help them think through a minimum viable idea they can ship this week. The interviewee should record and transcribe it. Here are some suggested questions — go off script when it helps.

  1. How are you feeling?
  2. What’s your idea?
  3. What’s the simplest version of your idea that someone might pay $1 for?
  4. Now re-think it in terms of outcomes: what wonderful result does your customer get for $1? Tip: people pay for results, not products or ideas.
  5. Why should they believe it will work?
  6. How quickly can they receive the value? And how easy can you make it for them? (Remember: time is money!)
  7. Would you pay $1 for that outcome?
  8. Who else do you think might want that outcome?
  9. How can I help?
  10. Is there anything else you’re worried about, or do you think this simple plan to [deliver results] for [$1] will just work?

the offer in one sentence

I am making [extremely simple product] for [the kind of person, like me] who [struggles with / delights in / loves / is frustrated by] [the painful moment I had]. It gives them [the result they want] for [the price]. I hope it can help :)

3. build a barely-good-enough version

Use the build prompt to make a minimally viable product. If the product needs a page, onboarding, and payment to work, build them into the product. If a direct message and payment link are enough for the first sale, use those.

Build to a Professional Bar Without Looping Forever →

Iterate on your product today and in the first half of the week if needed, then proceed to try to sell it.

4. make a public post about your product

Even if your product isn’t ready yet, you must share what you are building on a social platform of your choice, and ask people to DM you if they want to buy it early.

Tip: don’t worry if you’re afraid or shy. It’s commonly said that most entrepreneurs never get started because of two kinds of fear: fear of starting, and fear of asking. Remember that other people are kind-hearted, they want to see you succeed, and you are trying to help. Just ask!

Don’t worry about making it perfect — you can write another post tomorrow, and the next day. Post a screenshot of your post in the worksheet.

priya’s example post

I made a tool that helps me tweet more regularly. [screenshot] It’s a simple game where I compete with myself every day to post 1 tweet per day in each of 10 categories that are relevant to my life and work. Since making it I have had two tweets go viral (>5k likes) which is unusual for me. I still need to get a landing page, but I’m selling a prompt that will build you a custom one, just like mine, for $1 — just DM me or reply here for a payment link.

generic template

I made [thing] for myself because [problem it was solving]. It’s been really useful to me, for example [why it’s great]. If you’d like to buy your own I’m selling [thing] for [price]. [How to buy it right now.]

5. homework

Run this prompt to kick off your homework for the week:

Make a Plan →