📖 Sprint Storyboard

Broker sends a message to "Auction Central AI" on WhatsApp Business. The conversation looks like a normal WhatsApp chat. The broker types the property address, or sends a Property24 link. The AI responds: "Got it! 21 West Road, Morningside. I'll pull the data. In the meantime, what's the seller's name?"

AI asks 3–4 structured questions: seller name, reason for selling, preferred auction date, any special features. The broker responds naturally. AI confirms: "All set. Generating your proposal now — ready in ~3 minutes. I'll send you a link."

While the broker waits, the AI engine: (1) Pulls Lightstone data — comps, transfers, municipal valuation, (2) Scrapes P24 listing — photos, features, description, (3) Generates AI property description matching AC brand voice, (4) Calculates reserve recommendation from 5 closest comparable sales + suburb trend.

The broker sees a progress indicator: "Pulling property data... Analysing comparable sales... Writing description... Almost done."

WhatsApp message arrives: "Your proposal for 21 West Road, Morningside is ready! Review it here → [link]. Suggested reserve: R2.75M based on 5 comparable sales."

The broker taps the link → opens the proposal review dashboard in their browser.

The broker lands on the proposal preview screen — the core sprint test moment. They see the complete AI-generated proposal with sections: property overview, comparable sales, suburb trends, AI valuation, cover letter. Each section has an "Edit" button.

AI-written property description matching AC's brand voice. The broker reads through and checks: Does this sound like something AC would write? Are the facts correct? Would I be comfortable presenting this to a seller?

Interactive table of 5 closest comparable sales — sorted by distance from subject property. Shows: address, sale price, date, size, distance. Each comp has a map pin. The broker checks: Are these actually comparable? Would I have picked these same comps manually?

Median price graph with 12-month trend line and direction indicator. Shows: YoY price change, number of transactions, average days on market. Contextualises the valuation within market momentum.

This is the single most important frame in the entire storyboard.

The broker sees the AI-suggested reserve: "R2.75M — Based on 5 most similar properties (avg R2.8M, median R2.65M, 6% YoY increase in Morningside). Municipal valuation: R2.4M. Confidence: High (4 of 5 comps within 15% of subject)."

The reasoning chain is fully visible. The broker can see exactly why the AI chose this number. They can agree, override, or ask for more detail.

The broker adjusts the reserve from R2.75M to R2.9M and adds a note: "Strong street demand, recent neighbour sale at R3.1M not yet in system." The AI acknowledges the override but the system records both values — building the accuracy feedback loop.

AI-generated personalised cover letter addressed to the seller. Uses property-specific details and AC's brand voice. The broker reviews and can edit before finalising.

Broker clicks "Generate PDF". The system renders a complete, pixel-matched branded proposal PDF using the Handlebars template and proposal CSS. Output: a 7-section PDF identical in quality and branding to AC's current hand-crafted Word template — but generated in seconds.

Broker clicks "Share with Seller" → gets a unique tracked link. Can share via WhatsApp, email, or SMS. The link opens an interactive web version of the proposal (Qwilr-style) where the seller can explore maps, charts, and comps. Every view is tracked.

Broker checks the analytics: "Seller opened proposal 3 times. Spent 4 min on comparable sales, 2 min on valuation, 45 sec on cover letter. Last viewed: 2 hours ago." This gives the broker intelligence about what the seller cares about before the meeting.

The seller, after reviewing the interactive web proposal, clicks "Accept & Sign Mandate" at the bottom of the proposal. A digital signature flow captures their agreement to the 10-clause, 60-day auction mandate. No separate document. No printer. No in-person meeting required.

The deal is closed without leaving the proposal.