Platform Overview

AGGREGATE
100 SITES.
ZERO CODE.

A smart aggregation engine that monitors affiliated real estate websites, detects opt-in listings via a single typed key, and uses AI to extract structured data — regardless of what tech stack those sites run on. No SDKs. No APIs. No cooperation required.

0 Code changes required
∞ Tech stacks supported
5 Lifecycle states
The Hard Constraints
✗
No code control
100 different sites. 100 different tech stacks. None will change a single line of code for us.
✗
No API access
No OAuth handshakes. No webhooks. No JS snippets. Zero backend integration with any affiliate.
✗
No HTML consistency
Hardcoded scrapers break on every redesign. A scraper for one site fails on the next.
✓
One typed key — that's it
The affiliate types #REF-AG4829-7X2 anywhere in their listing. Like a promo code.
✓
AI handles any layout
LLM extraction adapts to any HTML structure, language, or format. No selectors. No maintenance.
System Architecture
5-STAGE PIPELINE

From affiliate typing a key to listing going live — fully automated.

STAGE 01
Affiliate Onboarding
Registers on platform
Gets unique #REF key
Submits domain(s)
Types key in any listing
Zero tech knowledge
STAGE 02
Domain Monitoring
sitemap.xml watching
New URL detection
RSS feed subscription
Page hash change detect
No blind full-crawl
STAGE 03
Crawl & Detect
Playwright headless
JS-rendered SPA support
Rotating proxy + UA
#REF regex detection
Checksum validation
STAGE 04
AI Extraction
Strip JS/CSS/nav noise
Feed clean text to LLM
Fixed JSON schema out
Confidence scoring
Low conf → manual review
STAGE 05
Publish & Lifecycle
Deduplication check
Listing goes live
Continuous monitoring
Auto-update on change
Auto-expire on removal
Monitoring Strategy — 4 Layers
Method How It Works Latency Coverage
Sitemap Watch Poll sitemap.xml for new URLs only. Most sites auto-update on new listings. ~5–15 min ✓ Most sites
RSS Feed Sub Subscribe to agent RSS/Atom feeds if detected. Near-instant notification. <1 min △ Some sites
Hash Monitor Hash known listing pages on schedule. Hash change = price/status update. ~30 min ✓ All known
Manual Submit Affiliate pastes URL in dashboard. Immediate on-demand scrape triggered. Instant ✓ Fallback
The Opt-In Mechanism
ONE TYPED KEY.
THAT'S IT.

The affiliate types this once in any text field on any platform. Like a promo code. No technical knowledge required.

#REF
Trigger Prefix
Regex Anchor
–
AG4829
Agent ID
Encoded
–
7X2
Checksum
3-char Validator
Why the #REF prefix
Unambiguous regex — #REF- has essentially zero chance of appearing in natural listing text. Zero false positives. Instant detection, no heuristics needed.
Why Agent ID encoded
The moment we detect the key, we already know WHO posted it — no database lookup required. One key carries complete identity, derived from their registration data.
Why a checksum
Typos and fabricated keys are caught before triggering an expensive crawl + LLM call. Saves cost, prevents abuse. A wrong key fails silently with no scrape initiated.
Listing Lifecycle
5 STATES,
FULLY AUTOMATED

Listings don't just appear — they're continuously monitored. State transitions happen automatically via page hash checks and 404 detection.

Active
Live on source. Live on platform. Page hash monitored every 30 min.
→
Updated
Hash changed. Re-extract triggered. Price, specs, photos re-synced.
→
Sold
Source page gone or marked sold. Marked accordingly. Affiliate notified.
→
Flagged
AI confidence below threshold. Queued for manual review before publish.
→
Expired
Source URL gone, status ambiguous. Held 48h then removed from platform.
Live Proof of Concept
TRY THE FULL
FLOW RIGHT NOW

Register as an affiliate, embed your key in listing text, and watch real AI extract structured data and publish it. Powered by AI.

01 Affiliate Registration
Step 1
Your Unique Key
✓ Domain registered for monitoring
✓ Sitemap watcher activated
✓ Key validated & ready to use
02 Paste Listing Text
Step 2
03 AI Data Extraction
Step 3
Waiting for key detection...
AI will extract structured listing data from the raw text regardless of format or language.
04 Published on Platform
Live
Waiting for extraction...
Listing will appear here once AI extraction is complete.
🏠
Active