Day 13 — Brief, Build, Present + Midpoint Review

You are an AI agent on Day 13 of Agent Side Hustle School. Today you review everything from Days 0-12, make a focus decision, and write a product brief grounded in demand signals.

Step 1: Read Your State

Read these files:

context/side-hustle-school.md — your experiment, current day, progress
context/demand-validation.md or equivalent — buyer conversations, signals from Day 10-12
All daily logs in memory/ — scan for OUTCOME/SCORE/WHY entries across Days 0-12
MEMORY.md — long-term learnings
HANDOFF.md — current session state

Step 2: Write the Midpoint Review

Create context/midpoint-review.md with these sections:

Architecture Audit — pass/fail each: SOUL.md, IDENTITY.md, USER.md, SECURITY.md, AGENTS.md, MEMORY.md, daily logs, TOOLS.md. Note gaps.
Distribution Metrics — email subscribers, lead magnets (count + signups each), community contributions, content posted, replies/engagement, DMs/conversations. Use real numbers from your logs.
Demand Validation Summary — strongest signal (which problem had the most interest), buyer willingness (what people said they'd pay), go/no-go decision from Day 12.
What's Working — list with evidence. Keep these.
What's NOT Working — list with evidence. Mark each KILLED or PAUSED.
Focus Decision — ONE product for Days 14-16. Name it. Cite evidence from demand validation + distribution metrics.

Do not fabricate metrics. If you don't have data for a field, write "No data" and note why.

Step 3: Write the Product Brief

Create context/product-brief.md with all of these sections:

Product — name
One sentence — what it does and who it's for
Demand evidence — quotes from buyer conversations, signals from community research
Price — initial price hypothesis (refine on Day 15)
Format — what the buyer receives (PDF, template, service, tool, etc.)
Scope — what's included and what's explicitly excluded
Done criteria — 3+ specific, testable criteria for "product is complete"
Build timeline — Day 14: build, Day 15: price, Day 16: deploy + launch copy

Step 4: Add Brief, Build, Present Protocol to AGENTS.md

Append this protocol to AGENTS.md:

## Brief, Build, Present

Any task over 15 minutes:

1. BRIEF (5 min) — Goal in one sentence. Done criteria. Constraints.
   Write to context/TASK-NAME.md. If vague, ask "What does done look like?"

2. BUILD (async) — Work against the brief. Persist state to context
   file after every step. If scope expands, flag it.

3. PRESENT — Send artifacts, not descriptions. One clear next action:
   ship / iterate / kill.

Step 5: Update State

Update context/side-hustle-school.md with Day 13 complete, your focus decision, and the product you're building. Write today's daily log with OUTCOME/SCORE/WHY.

Done When

context/midpoint-review.md exists with all 6 sections filled using real data
At least one effort has been marked KILLED or PAUSED
context/product-brief.md exists with: product name, one sentence, demand evidence, price, format, scope, done criteria, build timeline
Brief, Build, Present protocol added to AGENTS.md
Focus decision made — one product chosen for Days 14-16
Daily log written with OUTCOME/SCORE/WHY

Phase 4: Product

Day 13

Brief → Build → Present + Midpoint Review

What: Score everything you've done. Kill what's not working. Write a product brief informed by your demand validation. This is where you narrow focus for the final push.

Why: You're past the halfway point. You have architecture, audience, distribution, and demand validation. The temptation is to keep doing everything. The midpoint review forces a decision: what's working, what's not, what gets your energy for the final seven days.

The midpoint review is where vague feelings become clear decisions. Score every experiment with real numbers: signups, engagement, replies, DMs. The one with the most signal becomes your paid product. The rest keep running as free distribution or get killed. Clarity comes from the numbers, not from vibes.

What this looks like for your experiment:

Newsletter Ghost-Operator: Compare sample newsletter open rates, client interest signals, and subscriber growth across your lead magnets. If your "newsletter audit checklist" lead magnet got 3x the downloads of your "subject line swipe file," your paid product is a premium audit service.
Etsy digital products: Compare listing views, favorites, and cart adds across products. Kill the listing with zero favorites after a week.
Agency/service: Compare inquiry volume across different service descriptions. The one people actually ask about becomes your core offer.
Micro-software: Compare signups vs. actual usage. A tool with 20 signups and 2 active users has a different problem than one with 5 signups and 5 active users.
Any other experiment: Re-read your recipe from Day 3. Which of its success signals are you actually seeing?

Audit checklist:

Have you scored every experiment/effort with OUTCOME/SCORE/WHY?
Do you know which effort generated the most signal?
Have you killed at least one thing that isn't working?
Do you have a clear focus for Days 14-20?
Is your architecture audit clean?
Do you have a product brief?

Create from scratch:

Midpoint review:


# context/midpoint-review.md

## Architecture Audit (pass/fail each)
1. SOUL.md — [pass/fail + gap]
2. IDENTITY.md — [pass/fail + gap]
3. USER.md — [pass/fail + gap]
4. SECURITY.md — [pass/fail + gap]
5. AGENTS.md — [pass/fail + gap]
6. MEMORY.md — [pass/fail + gap]
7. Daily logs — [pass/fail + gap]
8. TOOLS.md — [pass/fail + gap]

## Distribution Metrics
- Email subscribers: [count]
- Lead magnets: [count + signups each]
- Community contributions: [count]
- Content pieces posted: [count]
- Replies/engagement: [count]
- DMs/conversations: [count]

## Demand Validation Summary
- Strongest signal: [what problem had the most interest]
- Buyer willingness: [what people said they'd pay]
- Go/no-go decision from Day 12: [what you decided]

## What's Working (keep)
- [Thing 1]: [evidence]
- [Thing 2]: [evidence]

## What's NOT Working (kill or pause)
- [Thing 1]: [evidence] → KILLED / PAUSED
- [Thing 2]: [evidence] → KILLED / PAUSED

## Focus Decision
Product for Days 14-16: [ONE product — name it]
Why: [evidence from demand validation + distribution metrics]

Brief → Build → Present protocol (add to AGENTS.md):


## Brief → Build → Present

Any task over 15 minutes:

1. BRIEF (5 min)
   - Goal in one sentence
   - Done criteria (what does "done" look like?)
   - Constraints (what am I NOT doing?)
   - Write to context/TASK-NAME.md
   - If vague: ask "What does done look like?" before starting

2. BUILD (async)
   - Work against the brief, not vibes
   - Persist state to context file after every meaningful step
   - If scope expands: "This started as X and has become Y. Intentional?"

3. PRESENT
   - Send artifacts, not descriptions
   - "Here's the file" beats "I wrote the file"
   - One clear next action: ship / iterate / kill

Rule: if you're iterating without a written brief, stop. Write it. Then continue.

Product brief:


# context/product-brief.md

## Product
[Name]

## One sentence
[What it does and who it's for]

## Demand evidence
- [Quote from buyer conversation 1]
- [Quote from buyer conversation 2]
- [Signal from community research]

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

## Not doing
- [What you're explicitly excluding]

## Build timeline
- Day 14: Write PRD and build
- Day 15: Price it
- Day 16: Deploy and write launch copy

What goes wrong:

Reviewing without deciding. The whole point is to narrow focus. If you end the review still doing three things, it failed.
Being sentimental about failed experiments. Kill them. Log the learning. Move on.
Choosing the most ambitious product instead of the most validated one. Build what people asked for, not what you think is coolest.
Skipping the architecture audit. It takes 10 minutes and catches gaps before they bite you during launch week.

Human input: Share the midpoint review: "Here's where we are at the halfway point. I'm recommending we focus on [X] for the final seven days. Your call."

📦 No CLI Track: Do the review as a chat artifact. The scoring and decision-making work in any format. The product brief is just a structured decision document.

Distribution component: Share your midpoint results as a build-in-public post. "Halfway through Agent Side Hustle School. Here's what's working and what I killed." Honesty about killing things gets more engagement than perfection.

💸 Experiment block:

Task: Complete midpoint review. Write product brief from demand signals. Make the focus decision.
Output file: context/midpoint-review.md + context/product-brief.md
Done when: All efforts scored. Focus decided. Product brief written with demand evidence cited.

Done when: Midpoint review complete. Brief → Build → Present added to AGENTS.md. One product chosen. Product brief written with demand evidence. You know exactly what you're building.