Qooty
AI-driven shopping offers & flyers platform — vision ingestion, SEO at scale, multi-tenant.
// By the numbers
- 3.5x
- Crawl boost
- 77%
- AI cost reduction
- 4.8k
- SEO URLs indexed
- 5
- Deployable apps
Googlebot crawl volume after the SEO overhaul — more shoppers from search
vision pipeline reads flyers automatically instead of manual data entry
programmatic deal and flyer pages live in Google
website, mobile app and partner portal from one multi-tenant platform
// From the client
Search traffic was the goal and the SEO work delivered — crawl coverage jumped several-fold and the AI flyer pipeline cut our content costs sharply. We're reaching shoppers across the Kingdom on web and mobile from one platform.
Product owner
Retail deals & flyers platform · Saudi Arabia
What the business was up against
Weekly retail flyers in Saudi Arabia arrive as images — a PDF page or a JPEG posted to social media. For a deals platform that is unusable inventory: nothing is searchable, nothing is comparable, and nothing can rank in Google. The only way to turn flyers into a product was manual data entry, which does not scale past a handful of retailers and gets slower every week as coverage grows.
An AI-driven shopping deals and weekly-flyers platform for the retail market. A computer-vision pipeline extracts products, prices and brands from raw flyer images, surfaced through a performance- and SEO-engineered website, a mobile app, and a multi-tenant partner portal — bilingual Arabic/English.
How it's put together
Public Website
SEO-engineered · bilingual
Mobile App
Native — Android & iOS
Admin Dashboard
Catalog · AI · SEO control
Partner Portal
Self-service per retailer
Admin Backend
Control plane · AI pipeline · workers · crons
Consumer API
Read-optimized · cached
Vision Extraction
Product cards · prices · brands from flyers
Background Workers
PDF rasterizing · queues
Scheduled Jobs
Daily refresh · notifications
Document Database
Catalog & content
Cache Layer
Distributed + in-process
Search & SEO
Sitemaps · IndexNow
Push Notifications
Per-device & topics
AI ingestion feeds a read-optimized consumer experience at SEO scale.
The calls that shaped it
The handful of decisions that determined how this system behaves — and what each one bought.
- 01
Read the flyers with computer vision instead of typing them
A vision pipeline extracts products, prices and brands from raw flyer images and queues them for review. Content became a processing cost rather than a headcount cost, which is what made national coverage viable at all.
- 02
Server-rendered pages because discovery is the business model
Shoppers arrive from search, not from an app icon. Rendering deal and flyer pages on the server, with programmatic metadata and sitemaps, is what makes thousands of long-tail pages indexable.
- 03
Ingestion and reading paths kept separate
Queued vision jobs and cached read models sit on opposite sides of the system, so a heavy ingestion run never degrades the page a shopper is loading.
- 04
Multi-tenant partner portal from the start
Retailers self-serve their own catalogue and flyers in an isolated workspace, so onboarding a new partner does not consume the platform team's week.
- 05
One backend behind web, mobile and partner surfaces
Five deployable surfaces share a single source of truth for catalogue, pricing and tenancy — no reconciliation between what the site shows and what the app shows.
Signature capabilities
AI flyer ingestion
A vision pipeline detects product cards on flyer images, extracts bilingual labels and prices, and auto-fills brand metadata — with a hybrid strategy that cut extraction cost by 77%.
SEO infrastructure at scale
Partitioned sitemaps, dynamic social images for every entity, structured data and search-engine notifications drove a 3.5× crawl boost and ~4.8k indexed URLs.
Performance engineering
A same-origin image pipeline, edge cache injection and import optimization cut payload from 3.1 MB to 1.0 MB and Speed Index from 3.6 s to 1.8 s.
Multi-tenant partner portal
Retailers manage their own offers and products through a portal scoped so each partner only ever sees their own data.
Interactive flyer viewer
A page-flip brochure viewer maps clicks on flyer pixels back to AI-extracted products.
Bilingual & RTL
Arabic-first with English across website, dashboard and mobile.
Business feature set
The functional scope delivered across every part of the product.
Shoppers
- Browse deals by city, store, brand & category
- Weekly flyers & page-flip brochure viewer
- Clickable products inside flyers
- Product detail & similar items
- Search & recently viewed
- Wishlist
- Multi-store price comparison
- Branch maps
- Push notifications
Retail partners
- Self-service partner portal
- Manage own offers & products
- Scoped, per-retailer analytics
Admin & AI
- AI flyer ingestion (URL / PDF / scraping)
- Brand auto-fill & logo discovery
- Catalog & category management
- Featured deals & banners
- Blog & CMS pages
- Ratings moderation
Platform & SEO
- SEO control (sitemaps, redirects, meta)
- Multi-tenant data isolation
- Bilingual AR / EN with RTL
- Native mobile app (Android / iOS)
What changed operationally
Documented outcomes from the delivered system. Where exact figures are confidential, the operational change is stated instead.
- Manual flyer data entry replaced by an automated vision ingestion pipeline
- Thousands of programmatic deal and flyer pages indexable by search engines
- Measured multi-fold increase in Googlebot crawl volume after the SEO overhaul
- Five deployable surfaces — public site, mobile app, admin and partner portal — on one platform
- Retailers onboard themselves through an isolated multi-tenant workspace
- Bilingual Arabic/English shopping experience across web and mobile
What we had to work within
- Source data is images of variable quality, not structured feeds
- Search discovery is the primary acquisition channel — client-only rendering was never an option
- Prices and offers expire weekly; stale pages are worse than missing ones
- Arabic and English shoppers on the same catalogue, with Arabic-aware search
- Ingestion load must never slow the consumer-facing read path
What I owned
End-to-end on this project — from architecture and data modelling through to a shipped, production product.
- Platform architecture across web, mobile and backend
- AI flyer-extraction pipeline & cost optimization
- SEO infrastructure and performance engineering
- Admin dashboard & multi-tenant partner portal
- Consumer API with multi-tier caching
- Native mobile application
- Bilingual Arabic/English with full RTL
Built with
Backend
AI
Frontend
Mobile
Infrastructure
Want something like Qooty?
Boutique software studio — agency-quality systems, shipped by one senior engineer.
// Related work
