Notes7 min read

Write the press release before you build

Write the launch press release and FAQ first. If you can't state the customer benefit plainly in a short release, or the hard questions have no good answers, don't build it yet.

From the course · lesson 1.8

With an AI agent, you can build a working version of almost any idea in a weekend. That changes where the risk sits, and it makes judgment the hard part. The expensive mistake is no longer writing bad code. It's spending a month of evenings on something nobody needed. Knowing whether an idea is worth building is now the real work.

So before you build, write the announcement. Pretend the product has launched and write the press release that tells the world about it. Then write the questions people would ask, and answer them honestly. If the release is dull or vague, or the answers fall apart, you've learned that for the price of an afternoon.

This is how lesson 1.8 of the course works. Here's the method, a template, and a short example using Kudos, a made-up testimonial app.

How to know an idea is worth building: Amazon's PR/FAQ

Amazon calls this "working backwards." You start from the customer and work back to the product, instead of starting from what you can build. The document is called a PR/FAQ: a press release plus frequently asked questions. Colin Bryar and Bill Carr, who both worked at Amazon, describe it as a "forcing function" that keeps the author focused on the customer.

The press release has a few fixed parts:

  • Headline and subheading. The product name, then one sentence on who it's for and what they get. Great products serve a specific group of customers, so name them.
  • Summary paragraph. Where and when it launches, and a short overview of the product and its main benefit. Pick a real launch date you expect to hit.
  • Problem. The pain, described from the customer's side, not yours.
  • Solution. How the product fixes it, and why what people use today falls short.
  • Two quotes. One from you, and one from a hypothetical customer describing the benefit in their own words.
  • Getting started. How someone would try it today.

Then come two FAQs. The external FAQ answers what customers and press would ask: price, how it works, where to get it, how to get help. The internal FAQ answers what a tough reviewer would ask: what already solves this, what's hard to build, what it costs, what has to be true for it to work, and what could kill it.

At Amazon, many PR/FAQs never get approved, and Working Backwards lists the usual reasons. The idea is a copycat. The market is too small. The cost is high and the payoff is risky. Something in the FAQ is still unsolved. Each of those is a cheap thing to learn on paper and an expensive thing to learn after launch.

The skeptic's internal FAQ

The internal FAQ is the part people skip, because it's the part that hurts. When you've just written a glowing press release, you're the worst person to attack it.

So hand that job to someone else. In Claude Code, you can set up a separate agent (a subagent, which runs with its own instructions) whose only job is to find reasons not to build. Give it your press release and nothing else. Tell it to write the internal FAQ as a tough, fair reviewer would: specific questions, no compliments, and a note on which answers are guesses.

For Kudos, a skeptic might ask:

  • Freelancers already paste client emails onto their sites. Why would they pay to stop doing that?
  • Clients are busy. What makes them actually reply to a testimonial request?
  • Who writes the final quote, and how do we know the client approved it?
  • How many freelancers would pay monthly for this, and how would we reach the first fifty?

The last question is the one many builders leave too late. We dig into it in why AI-built apps get no customers. You answer each one yourself. Where you don't know, write "unknown" and turn it into something to test this week. A short list of honest unknowns is a good result.

The 11-star exercise

The press release says what you'll build. The 11-star exercise helps you decide how good the key moment needs to be.

Brian Chesky, Airbnb's co-founder, described it on Tim Ferriss's podcast. Most products aim for five stars: it worked. So keep going. A six-star stay has a host who welcomes you and a stocked fridge. Seven stars, they've lined up a surfboard and lessons because they know you surf. By ten stars, the Beatles are checking in and crowds are cheering. At eleven, Elon Musk is at the door to take you to space.

Nobody builds eleven stars. Chesky's point is that somewhere between "the door opened" and "I went to space" there's a sweet spot you would never have found by aiming for five.

Here's the exercise on Kudos's key moment, asking a happy client for a testimonial:

StarsWhat happens
1A generic "please review us" email arrives months later. The link is broken.
5A clean form arrives. It works.
6The form already knows the project name and takes a minute on a phone.
7It arrives the day the project ships, and the client can just talk while Kudos drafts a quote for them to approve.
9The client gets a lovely thank-you card with their words, ready to share.
11Their quote lands in a magazine profile of the designer.

The six and seven rows become your design target. They're specific, buildable, and better than what anyone expects.

Idea filterStandby
Many ideas go in, the press release and skeptic's questions cut most of them, and one scored idea comes out.

A template

Copy this into a file called prfaq.md and fill in the brackets. Keep the press release to about a page. If it needs more, the idea probably isn't clear yet.

# [Product name]: [benefit in under ten words]
[One sentence: who it's for and what they get.]

[City], [launch date]. Today [company] launched [product], which helps
[specific customer] [do what] so they can [result].

## The problem
[Two or three sentences in the customer's words. What hurts today?]

## The solution
[How it works, in plain words. What people use today, and why it falls short.]

## Quotes
"[Why we built it.]" said [your name], [role].
"[Imaginary customer describing the benefit.]" said [type of customer].
(Internal only. Never publish this quote.)

## Getting started
[What someone does today to try it, and what it costs.]

---

# External FAQ
1. How much does it cost?
2. How does it work?
3. Where do I get it?
4. How do I get help?

# Internal FAQ (written by the skeptic agent, answered by you)
1. What do customers use today, and why would they switch?
2. What is hardest to build?
3. What does it cost us per customer?
4. What must be true for this to work?
5. What are the top three ways this fails?
6. Any legal or platform risks?

# 11-star moment
Key moment: [the one moment that matters most]
1 / 5 / 6 / 7 / 9 / 11 stars: [one line each]
Build target: [the 6 or 7 star version]

Filled in briefly for Kudos, the top of it reads:

Kudos: your clients' praise, ready to show. For freelance designers who lose good feedback in email threads. Kudos asks clients for a testimonial the day a project ships, drafts the quote from what they say, and adds it to a wall on your site once they approve it.

The skeptic's first question cuts straight to it: "Why not just screenshot the email?" If your answer is weak, that's the thing to test first, before any code.

Try it

  1. Pick one idea you're excited about. Fill in the template in thirty minutes or less.
  2. Open Claude Code and ask a fresh subagent to act as the skeptic. Give it only your press release.
  3. Answer every question. Mark guesses as guesses.
  4. Run the 11-star exercise on your key moment and circle the six or seven.
  5. Decide: build, change, or drop. All three are good outcomes.

In the course, this becomes two reusable commands, /prfaq and /11-star (the same idea as a Claude Code skill), so you can run it on every new idea. You can see where it fits in the journey on the course page.

Sources

Learn this inside Claude Code.

Join the waitlist for Chief of Agents. You'll get the free beginner's handbook and skills right away, and hear first when the course opens.

One email to confirm, then the handbook and skills. Unsubscribe anytime.

Keep reading

  1. 6 min read

    Your first Claude Code skill in 15 minutes

    How to make a Claude Code skill: pick a weekly chore, write it down once as a SKILL.md, and run it any time by typing its name.

  2. 7 min read

    Write your taste down: a DESIGN.md your agents follow

    How to get AI to follow your design taste: write a DESIGN.md of short, checkable rules your agents read, then review screens against it.

  3. 6 min read

    Claude Code for designers: what it is, and why it isn't just for engineers

    What a coding agent does, why directing one is a design skill, and a safe first hour for designers new to Claude Code.

All notes →

© 2026 Chief of Agents · Notes · PrivacyIdeate. Build. Sell. You lead, the agents work.