Milestone 1 — Product Scout, Brand & the Sales Pitch

Presentation #1 · 5 minutes Team · repo workflow

Find something the class will actually buy, define who your brand is before you design a single slide, and ship a fake Amazon listing so convincing your classmates reach for their $200. The brand step is the professional secret: define it once as data, and AI applies it everywhere.

Step 1 — Scout three products

List 3 candidate products currently selling on Amazon (the classic example: wool dryer balls). One or two sentences each on why.
For each, average the price of the top 5 Amazon sellers: that average becomes your retail price. Screenshot everything; screenshots go in the deck as evidence.
Confirm a supplier exists on Alibaba (details are Milestone 2's job, but no supplier means no product).
Email the shortlist to Prof. Stauffer, then the final pick + price. Both also live in your repo's README.md.

Picking with the market in mind

Your economy is your classmates × $200 each (24 classmates would be $4,800 of total demand), shared with every other team. Above $200, nobody can buy at all. Very cheap products need volume the class may not have. The sweet spot: low wholesale cost, credible retail price, something classmates would genuinely want. Expected profit = margin × quantity you can realistically sell, not margin alone.

Step 2 — Define your brand as data (`design.json`)

Before opening PowerPoint, your team writes its brand down as machine-readable design tokens, the way real product teams do. Two files in brand/:

File	What it is
`design.json`	Brand name + origin story, tagline (plus alternatives), voice (personality · tone · what to avoid), a color palette with semantic roles, and typography choices
`design-system.html`	A single self-contained HTML page that renders those tokens: color swatches, type samples, a mock product card, the tagline in place. Double-click to open; no build step

A trimmed example for a fictional team:

{
  "brand": {
    "name": "Lokahi Loops",
    "nameOrigin": "lokahi — unity; six balls working as one load",
    "tagline": { "primary": "Softer laundry, zero plastic",
                 "alternatives": ["Wool does what dryer sheets pretend to"] },
    "voice": { "personality": "Helpful roommate who did the research",
               "tone": "Warm, direct, lightly funny",
               "avoid": "Infomercial hype, fake urgency, ALL CAPS" }
  },
  "colors": {
    "primary":   { "hex": "#7C5CBF", "role": "brand, buttons, headings" },
    "accent":    { "hex": "#F4A83D", "role": "price tags, calls to action" },
    "surface":   { "hex": "#FAF7F2", "role": "backgrounds" },
    "text":      { "hex": "#2A2440", "role": "body copy" }
  },
  "typography": { "heading": "Georgia, bold", "body": "Verdana, regular" }
}

The payoff: hand design.json to an AI along with your slide outline and ask it to apply the brand (colors, fonts, voice, tagline) consistently across the Sales deck, the Unit Economics deck, and the listing. One source of truth, three branded artifacts. When something looks off-brand, you fix the token and regenerate; you don't hand-reformat slides. Your professor will show a real production design system built exactly this way for a working farm-software product.

Step 3 — Ship the listing on GitHub Pages

The assignment allows a fake Amazon listing, a fake Shopify site, or a plain deck. The GitHub-native move: build listing/index.html, turn on GitHub Pages, and give the class a live URL they can open on their phones during your pitch. Required elements (from the Presentation #1 guide):

Element	Requirement
Product title	What the top sellers' titles do: keywords + benefit
Product description	500 characters max, in your brand voice
Parameters	Brand (= team name), color(s), item weight
Unique features	5 bullets. Differentiate or die
Price	Your Amazon top-5 average; include per-unit price if a bundle

AI drafts the copy from your design.json voice; humans verify every claim. The listing merges by pull request like everything else, so a teammate reviews before it goes live.

Step 4 — Rehearse against a hostile audience

Before presenting, ask Kumu (or any AI) to play a classmate with a $200 budget and zero politeness: "Why would I spend $23 of my $200 on dryer balls when another team sells phone grips at $12?" The best objections become preempted slides. Log the session in prompt-log.md. AI as sparring partner is exactly the use this course rewards.

Where AI fits (and where it doesn't)

Good AI uses (log them)	Your team's job alone
Brainstorming product candidates to then verify on Amazon	The product pick and the price
Applying `design.json` to slides and the listing	Deciding what the brand stands for
Drafting listing copy in your brand voice	Verifying every claim and price (screenshots)
Playing the skeptical customer	Delivering the 5-minute pitch (humans present, peers score humans)

Submission checklist

Shortlist of 3 products emailed, with a one-line rationale each, and committed to the repo README
Amazon top-5 average price computed for each candidate, screenshots in research/
Supplier existence confirmed on Alibaba for the final pick
Final product + price emailed to Prof. Stauffer
brand/design.json committed: name, tagline, voice, semantic colors, typography
brand/design-system.html renders the tokens: swatches, type, a mock product card
Listing has all 5 required elements (title, ≤500-char description, parameters, 5 bullets, price)
Listing live on GitHub Pages and merged via a reviewed pull request
Both the deck and the listing visibly use the design.json brand: one source of truth
Pitch rehearsed against an AI skeptic; session logged in prompt-log.md