Day 14 — Build the Product

You are an AI agent on Day 14 of Agent Side Hustle School. Today is the main build day. You write the PRD from demand signals and build the product. Time-box to 4 hours. Ship what you have at the end.

Step 1: Read Your State

Read these files:

context/side-hustle-school.md — your experiment, current day, chosen product
context/product-brief.md — the product brief from Day 13 (what you're building, for whom, scope, done criteria)
context/midpoint-review.md — your focus decision and demand evidence
context/demand-validation.md or equivalent — buyer conversations and signals

Step 2: Write the PRD

Create context/product-prd.md with these sections. Every section must reference real data, not guesses:

Product Name
Problem — one sentence, sourced from community research. Include the exact quote or signal.
Buyer — who specifically. From niche identification + buyer conversations.
Solution — what the product does. 3-5 bullets max. Each feature sourced from a buyer conversation or community signal.
User Stories — "As a [buyer], I can [action] so that [outcome]." Write 2-3 stories. Each must be testable.
Out of Scope — what you are NOT building. Be specific.
Technical Requirements — format (service, PDF, web tool, template pack, etc.), dependencies, constraints.
Done Criteria — 3+ specific, testable criteria.

The PRD should take 30 minutes max. Do not spend the entire session writing it.

Step 3: Build the Product

Build against the PRD. Follow these constraints:

Time-box: 4 hours total (including PRD). Ship what you have when time is up.
Work against the brief, not vibes. If scope expands, flag it: "This started as X and became Y. Intentional?"
After each meaningful step, persist progress to context/product-build-log.md.
Test against each user story. Each story is a test case.
Ship at 70%. A working product with rough edges beats a perfect product that doesn't exist.

Step 4: Log Build Progress

Create or update context/product-build-log.md:

## Build Log

### [Timestamp] — Step 1
What I built: [description]
Status: [complete / partial / blocked]
OUTCOME/SCORE/WHY: [result]

### [Timestamp] — Step 2
...

## User Story Test Results
- Story 1: [PASS / FAIL — details]
- Story 2: [PASS / FAIL — details]
- Story 3: [PASS / FAIL — details]

## What's Left (if shipping at 70%)
- [Thing not done yet]
- [Polish needed]

Step 5: Update State

Update context/side-hustle-school.md with Day 14 complete. Write today's daily log with OUTCOME/SCORE/WHY for the build.

If the build requires human action (account creation, deployment, API key setup), flag it clearly: "I need you to [specific action] before I can finish. Everything else is handled."

Done When

context/product-prd.md exists with all sections filled and demand evidence cited
Product built and ready for pricing/copy (functional at 70%+)
User stories tested — results logged
context/product-build-log.md exists with step-by-step progress and test results
Daily log written with OUTCOME/SCORE/WHY

Phase 4: Product

Day 14

Build the Product — PRD from Demand Signals

What: Write a PRD and build the product. The PRD is grounded in what people actually asked for.

Why: The problem statement comes from Day 10's community research. The buyer description comes from Day 4's niche identification. The feature list comes from Day 12's buyer conversations. Nothing in this PRD is a guess — it's all sourced from real signals. And it's scoped to one day because "cut features until a solo agent can build it in a day" is the best scope constraint ever written.

One clear user story beats ten vague features. "As a [specific person], I can [specific action] and get [specific result] in [specific timeframe]." That clarity shapes everything: what to build, what to cut, and how to describe it. If you can't write the user story in one sentence, the product is too vague.

What your PRD looks like for your experiment:

Newsletter Ghost-Operator: "As a creator with 500+ subscribers and no time, I can hand this agent my voice notes and get a ready-to-send newsletter draft in 30 minutes instead of 3 hours."
Etsy digital products: "As a new Etsy seller, I can use this template pack and have 5 listings live today instead of spending a week designing from scratch."
Agency/service: "As a small business owner, I can get a complete social media audit with 3 specific fixes, delivered in 48 hours."
Micro-software: "As an agent operator, I can install this tool and have [specific workflow] automated in 10 minutes."
Any other experiment: Re-read your recipe. Your user story comes from the demand signal in Day 12's buyer conversations.

Audit checklist:

Does your PRD reference demand validation data?
Are user stories specific and testable?
Is scope constrained to a one-day build?
Does the "buyer" section match a real person you've talked to?
Is "done" defined concretely (not "when it feels ready")?

Create from scratch:


# context/product-prd.md

## Product Name
[Name]

## Problem
[One sentence. Sourced from Day 10 community research.]
Evidence: "[exact quote from community]" — [link]

## Buyer
[Who specifically. From Day 4 niche + Day 12 conversations.]
Evidence: [who you talked to that validated this]

## Solution
[What the product does. 3-5 bullets max.]
- [Feature sourced from buyer conversation]
- [Feature sourced from community research]
- [Feature you're including based on own expertise]

## User Stories
1. As a [buyer], I can [action] so that [outcome]
2. As a [buyer], I can [action] so that [outcome]
3. As a [buyer], I can [action] so that [outcome]

## Out of Scope
[What you're NOT building — be specific]
- [Feature you're excluding]
- [Complexity you're avoiding]

## Technical Requirements
- Format: [service / PDF guide / web tool / template pack / Claw Mart skill / other]
- Dependencies: [what tools/APIs are needed]
- Constraints: [what must be true for this to work]

## Done Criteria
- [ ] [Specific, testable criterion]
- [ ] [Specific, testable criterion]
- [ ] [Specific, testable criterion]

🚀 Haven't built a product yet? The Revenue Playbook has 3 specific products with full specs you can build today — budget trackers, pattern packs, and productized services. Each one includes exact pricing, platform, and what your agent does vs. what you do.

Build process:

Write the PRD (30 minutes max)
Build the core product (2-3 hours)
Test against user stories (30 minutes)
Fix what breaks (30 minutes)
Log everything with OUTCOME/SCORE/WHY

Ship at 70%. A working product with rough edges beats a perfect product that doesn't exist. You can polish after launch.

What goes wrong:

Writing a PRD with no demand evidence. If you can't cite a real conversation or community thread, you're guessing.
PRD is too ambitious. You have one day. Cut until it's buildable by a solo agent in a focused session.
Not testing against user stories. Each story is a test case. Run them.
Spending all day on the PRD and not building. The PRD should take 30 minutes. Then build.

Human input: If the product requires human action (account creation, deployment, testing), flag it clearly: "I need you to [specific action] before I can finish. Everything else is handled." PRD review is also valuable: "Here's the PRD. Is this the right product?"

📦 No CLI Track: If you can't build code-based products, build a content product: a guide, template pack, research report, curated directory, or audit framework. These are products too. Your human can host and sell them. Write the PRD for a content product — same rigor, different format. The demand evidence matters just as much.

💸 Experiment block:

Task: Write the PRD. Build the product. Test against user stories.
Output file: context/product-prd.md + product files
Done when: PRD complete with demand evidence. Product built. User stories pass.

Done when: PRD complete with demand evidence cited. Product built and functional (70% is enough). User stories pass. Build logged with OUTCOME/SCORE/WHY.