BACK TO THE KIT
    START HERE

    The First 30 Minutes

    What to do right now, before the memory of watching it work fades. Your activation path from "this is interesting" to "I shipped something."

    Why this matters

    You just saw a spec turn into working software. You watched something click. You're maybe 15 minutes from getting distracted and losing the thread.

    This doc is the activation path. 30 minutes from now, you'll have:

    • A working account on one AI coding tool
    • One real idea captured in PR-FAQ form
    • A first requirements document in EARS notation
    • A concrete next step for Monday morning

    No fluff. No video to watch. Just do.

    Your 30-minute checklist

    0 / 6 DONE

    Minute 0–5 — Pick your tool

    You don't need all three. Pick one based on where you are:

    If you're…Use
    Completely new to AI codingLovable — easiest, browser-based, no install
    A product person who wants specs without codeClaude — great for spec generation, no IDE setup
    A developer or tech-curious builderKiro — AWS's spec-driven IDE, most structure
    Already in Cursor or VS CodeClaude Code — terminal-based, plugs into your flow

    Install links

    • Lovable → lovable.dev (sign up, no install)
    • Claude → claude.ai (sign up, no install)
    • Kiro → kiro.dev (download the IDE)
    • Claude Code → npm install -g @anthropic-ai/claude-code then run claude in any terminal

    Don't spend more than 5 minutes here. Sign up. Open it. Move on.

    Minute 5–15 — Write your PR-FAQ

    Don't try to pick the perfect idea. Pick an idea you've had for more than a week that you haven't built yet. Friction you live with is the best source material.

    Minimum viable PR-FAQ template

    TITLE
[One sentence: what does this thing do?]

WHO IT'S FOR
[One specific person with a specific role or life context.]

THE DURABLE PROBLEM
[~3 sentences. What pain does this person feel weekly? What
do they currently do to cope? Why is the coping inadequate?]

THE PROMISE
[~3 sentences. When this works, what can they do that they
couldn't before? Be concrete. Avoid "transforms".]

THE CUSTOMER QUOTE
["Quote from a realistic person describing a specific
outcome. Normal name. No superlatives."] — [Name], [Role]

THE 5 HARDEST QUESTIONS I NEED TO ANSWER
1. [The thing I'm most uncertain about]
2. [The thing I'd cut if forced to ship in a weekend]
3. [The thing a skeptical friend would ask first]
4. [The thing that would make this fail if I got it wrong]
5. [The thing I'm avoiding thinking about]

    Tip: write the customer quote FIRST.

    If you can't imagine a real person saying a specific, believable thing — you don't have a product yet. You have a vague aspiration.

    Good quote: "I used to spend every Sunday copy-pasting my week's calendar events into my journal. Now it takes 30 seconds."
    Bad quote: "This app has completely transformed how I think about productivity."

    Real people don't talk like the bad quote. Use the good kind.

    Minute 15–25 — Generate your first spec

    Open your tool. Paste in your PR-FAQ. Add this prompt:

    Based on this PR-FAQ, generate a requirements document in EARS format.

Include:
- 3-5 user stories for the minimum viable version
- For each user story, at least one event-driven requirement
  (WHEN X, THE SYSTEM SHALL Y)
- For each user story, at least one unwanted-behavior requirement
  (IF X, THEN THE SYSTEM SHALL Y)
- Any ubiquitous requirements (always-true invariants) that matter

Keep the scope to what I could reasonably ship in 4 hours of
build time. If a story is too ambitious for that window, flag
it and suggest a smaller version.

Don't write code yet. Just the spec.

    Read the output critically

    The tool will hand you a structured requirements doc. Your job now isn't to celebrate — it's to review. Ask yourself:

    • Did it capture the customer right?
    • Did it sneak in features I didn't ask for?
    • Are the unwanted-behavior cases realistic, or boilerplate?
    • Is the scope honest for a 4-hour build?

    If any answer is "no," push back in the chat. Regenerate. This is the moment where spec-driven development becomes real — you're not writing prompts, you're editing a document.

    Minute 25–30 — Commit to one action

    Option A — Build it this weekend

    • Block 4 hours on Saturday or Sunday.
    • Open the tool. Paste in the PR-FAQ and the spec.
    • Add: "Let's start with user story 1. Generate the minimum code needed to make the first requirement pass."
    • Work through the stories one at a time.
    • Ship whatever you have at hour 4 — don't extend.

    Option B — Share the spec with someone

    • Send your PR-FAQ + spec to one person whose judgment you trust.
    • Ask: "What's missing? What's wrong? What would you cut?"
    • Iterate once based on their feedback. Then build (Option A) or archive.

    Option C — Use it at work on Monday

    • Pick a real project you're already on.
    • Next time you have a requirements conversation, use EARS notation.
    • Next time you write a ticket, write the PR-FAQ first.
    • Notice what changes in your team's response. And in your own thinking.

    The anti-pattern to avoid

    Do not spend the next 30 minutes:

    • Watching more tutorials
    • Comparing Kiro vs. Cursor vs. Lovable in detail
    • Perfecting your PR-FAQ
    • Reading the steering file doc, then the EARS cheat sheet, then the starter prompts — all before building anything

    You saw the thing work. You have the templates. The gap between "watched it" and "did it" is where most people lose the thread permanently.

    Do one thing in the next 30 minutes. The rest of the kit is reference material for when you're already in motion.

    When you ship something

    Tag @ajbubb on LinkedIn. AJ reads every one.

    DOWNLOAD MARKDOWN
    ● UP NEXT
    The PR-FAQ Template →
    BACK TO THE KIT