Abdelrhman
Refaat
I build AI and product systems that drive growth, reduce cost, and turn operational complexity into scalable advantage.
Nearly 4 years building
AI-powered products
in fast-scaling e-commerce
Product manager with nearly four years of experience building AI-powered products and operational systems across fast-scaling e-commerce platforms in the Middle East. Started in operations and worked my way up to leading the full AI product portfolio at Taager — shipping 13+ products that collectively save $155K+ annually, lift CVR by 15%, and automate workflows end-to-end. I think in systems, obsess over measurable outcomes, and ship.
Currently at Taager.com — a cross-border e-commerce platform scaling across the Middle East — as Lead AI Product Manager, owning the full AI product portfolio.
13+ Products Shipped.
Every one in production.
Click any highlighted card to open the full case study. All others show product details.
AI Click-to-WhatsApp Acquisition Agent
High cost-per-acquisition from paid social, manual follow-up, no scalable autonomous channel.
AI Recovery Agent for Failed Deliveries
Failed deliveries required expensive human call center agents to re-schedule and recover shipments.
AI WhatsApp Storefront Conversion Channel
Standard web storefront had high drop-off. Merchants needed a familiar, low-friction channel to browse and order.
Saudi National Address Validation Service
Address collection was manual, error-prone, expensive ($0.50/order), and a major driver of failed deliveries.
Self-Hosted Chatwoot Migration
High monthly spend on third-party customer support platform with limited customization.
AI Customer Support WhatsApp Agent
High volume of repetitive support queries consuming agent time and driving up support costs.
Autonomous Merchant for TikTok Campaigns
Managing TikTok ad campaigns manually was time-intensive and inconsistent across merchant accounts.
Prepaid Order Conversion Feature
High COD (cash on delivery) rate created cash flow risk and increased return rates.
Smart Reachability Reminder System
Failed delivery attempts due to unreachable customers were costly and drove redelivery overhead.
CX Reporting & C-SAT Measurement System
No structured way to measure customer satisfaction or surface CX trends to leadership.
Shipment Tracking Portal A/B Test
High inbound contacts about order status creating avoidable support volume.
Warehouse Scheduling Optimization
Inefficient warehouse scheduling causing overtime costs and delayed shipouts.
Analytics Pipelines & Operational Dashboards
Operations team lacked real-time visibility into key metrics — order flow, SLAs, delivery performance.
Cards with deep case studies open a dedicated page. Click any card for product details.
Deep dives on the
products that mattered most
Problem → decisions → results → what I learned. No fluff.
AI Recovery Agent: Turning Failed Deliveries into Recovered Revenue
Failed deliveries are expensive. Each one costs the business a re-delivery attempt, consumes call center agent time, and risks a permanent churn. We had a high-volume call center team whose primary job was calling customers after failed delivery attempts — confirming new addresses, rescheduling, and re-routing.
AI WhatsApp Storefront: 15% CVR Lift by Meeting Customers Where They Are
We had a standard web storefront for merchant acquisition. Conversion rates were flat. Merchants in MENA are deeply WhatsApp-native — it's how they communicate, shop, and do business. Asking them to navigate a web UI was friction we were introducing, not reducing.
Saudi Address API: 50x Cost Reduction Through Lateral Thinking
Every order in Saudi Arabia required collecting and validating a delivery address. Our manual collection process was expensive ($0.50/order), slow, error-prone, and a significant driver of failed deliveries — customers often provided incorrect or incomplete addresses.
Warehouse Scheduling: From Operator Insight to Product
As an Operations Specialist at Taager, I noticed our warehouse operated on a fixed shift schedule that was misaligned with actual order volume patterns. Monday mornings: overwhelmed, overtime, missed SLAs. Thursday afternoons: idle staff, wasted capacity. The mismatch was predictable and recurring.
How I approach the work
Six disciplines, practiced daily, grounded in real examples.
Strategy
Start with business pain and work backwards. Every roadmap item maps to a measurable outcome. I say no to more than I say yes to.
Discovery
Operator background means I go to the source. I've been in warehouses, listened to support calls, and sat with logistics teams before writing a PRD.
Data
I write my own SQL. I build my own dashboards. I don't wait for analysts to tell me what's happening — I go look.
Execution
PRD → engineering alignment → build → staged rollout → measurement. I stay in the details without micromanaging. Ship fast, measure rigorously.
Experimentation
Every significant feature gets an experiment design before it ships. Control groups, success metrics, and clear kill criteria defined upfront.
Scale
Build for 10x from day one. The AI recovery agent was designed to handle peak-season volume with zero marginal cost from the start.
Capabilities that
cross the full stack
How I think, decide, and ship.
Start with Pain, End with Proof
Every product I've shipped started with a specific, documented pain. Every product I've shipped has a specific, measured result. If you can't articulate the pain and you can't measure the outcome, you're not doing product — you're doing theater.
Think in Systems, Not Features
A feature solves one problem. A system solves a class of problems. When I built the AI recovery agent, I wasn't thinking about one failed delivery — I was thinking about every failed delivery, forever. That's the difference between a fix and a product.
Operator Insight is an Unfair Advantage
I started in operations. I've been in the warehouse, I've been on the support queue, I've listened to the recovery calls. When I design products for these workflows, I'm designing from lived experience — not from a distance. That's not something you can get from a PRD template.
AI is Infrastructure, Not a Feature
The mistake most companies make is bolting AI onto existing products. The right approach is asking: if AI existed from day one, how would we have built this? That question leads to fundamentally different — and better — products.
Simplicity is the Hardest Thing to Build
The best products I've shipped had fewer features than I originally scoped. Every feature you cut is a complexity you've saved your user from. The discipline is in what you don't build.
Strategy Without Execution is Fantasy
I've sat in rooms with brilliant strategies that never shipped. And I've seen mediocre ideas, executed perfectly, change businesses. I respect the strategy work — and I obsess over the execution work. You can't separate them.
Let's build something
worth shipping
Open to product leadership roles, AI product consulting, and interesting conversations about building in MENA.