AI Customer Service Automation UAE E-Commerce
How We Built an AI-Powered WhatsApp Customer Service System for UAE E-Commerce
Client
Eona.ae
Headquarters
Dubai, UAE
Timeline
3 months
Technology
WHATSAPP AI
About 90% of queries automated (client-reported)
70-80%
reduction in support costs (client-reported)
Replies in under 30 seconds (client-reported)
24/7
availability without 24/7 staff
The challenge
200+ daily support tickets, 4-hour response times, no after-hours coverage
The solution
Intelligent WhatsApp chatbot with smooth human handoff
The results
- About 90% of queries automated (client-reported)
- 70-80% reduction in support costs (client-reported)
- Replies in under 30 seconds (client-reported)
- 24/7 availability without 24/7 staff
Growing Pains in Customer Support
AI customer service automation for a UAE e-commerce brand. A WhatsApp bot now handles about 90% of support queries with replies in under 30 seconds, cutting support costs by 70-80%, as reported by Eona.ae. Eona.ae was experiencing explosive growth in the competitive UAE e-commerce and delivery market, but that growth came with a critical bottleneck in customer support.
The support team was drowning: 200+ tickets flooding in daily, average response times over 4 hours, and coverage only from 9 AM to 5 PM Dubai time while customers shopped and ordered around the clock. Pre-purchase questions went unanswered and carts were abandoned, customers waited on delivery updates, and the support tools had no connection to delivery tracking.
The business impact was clear. Slow responses were driving customers to competitors, the team's workload grew with every order, and after-hours shoppers had no way to get help. Agents answered the same questions hundreds of times, and customers preferred WhatsApp while support ran on email.
Traditional solutions fell short. Generic chatbot platforms could not handle complex delivery queries, hiring more agents would only add cost, and off-the-shelf tools had no integration with the custom delivery system. Eona.ae needed a purpose-built AI solution, integrated with their existing systems and deployed fast enough to capture the growth opportunity. That is when they reached out to EdgeFirm.
4 capabilities carried the build
We designed an intelligent WhatsApp chatbot that handles the majority of customer interactions and hands complex cases to human agents inside a single conversation thread. Four core capabilities carried the build.
01
Intelligent customer service automation
Natural language understanding in English and Arabic, dynamic response templates for FAQs, product info, and order status, context-aware conversations that remember previous interactions, graceful handling of ambiguous questions, and sentiment analysis that detects frustrated customers and escalates proactively
02
Real-time delivery integration
Live connection to Eona.ae's order management system, automatic delivery status updates, tracking number lookup with estimated delivery times, address changes and rescheduling, and proof-of-delivery notifications with photo confirmation
03
Smooth human handoff
Automatic escalation when bot confidence is low, agent notifications via dashboard, full conversation history preserved during handoff, mid-conversation takeover the customer never notices, suggested responses for the agent, and a clean hand-back to the bot once the issue is resolved
04
Backend dashboard for the support team
Real-time conversation monitoring, queue management for escalated tickets, performance analytics covering resolution time, satisfaction, and automation rate, conversation tagging and categorization, and historical conversation search
The stack behind it
WhatsApp Business API, Python + FastAPI, MongoDB, AWS EC2, Custom delivery system integration
AWS
RedisFrontend (Customer-Facing)
WhatsApp Business API for message handling, webhook integration for real-time message receipt, and media handling for customer photos and documents
AI/ML Layer
Custom NLP model for intent classification, context management for multi-turn conversations, sentiment analysis for escalation triggers, and confidence scoring for handoff decisions
Backend Services
FastAPI for high-performance API handling, MongoDB for conversation history and context, Redis for session management and caching, and Celery for background task processing
Integrations
Eona.ae order management system REST API, delivery tracking provider webhooks, real-time agent dashboard updates via WebSockets, and an analytics pipeline to the data warehouse
Deployment
AWS EC2 hosting for the bot services
Our development process
Weeks 1 to 2
Discovery and planning
We analyzed 500+ historical support tickets to identify patterns, mapped conversation flows for common scenarios, defined escalation triggers and handoff protocols, and integrated with Eona.ae's staging environment for testing.
Weeks 3 to 6
MVP development
We built the core chatbot around the top 10 most common queries, integrated the delivery API for basic tracking, created a simple dashboard for agents, tested with 5 internal team members, and iterated on real conversation feedback.
Weeks 7 to 10
Full feature rollout
The bot expanded to 50+ query types, gained Arabic language support and advanced sentiment analysis, and shipped with a full agent dashboard plus monitoring and alerting systems.
Weeks 11 to 12
Launch and optimization
A soft launch with 10% of traffic let us fix edge cases before training the support team on the new workflow, rolling out to all customers, and optimizing on real usage data.
Transparency
Transparency ran through the whole engagement. Every week Eona.ae saw what we built in a live demo, what we learned from testing, the coming week's priorities, and any blockers or risks with proposed solutions. No surprises, just clear, continuous visibility.
After launch
Post-launch, the engagement included 30 days of support with weekly check-ins, bug fixes and minor tweaks as issues arose, full documentation and runbooks, and training for Eona.ae's internal dev team. Eona.ae chose to continue on a monthly retainer for ongoing feature additions and performance monitoring.
Next steps
The roadmap keeps growing. Phase 2 adds voice note support with speech-to-text, in-WhatsApp payments for faster checkout, proactive messages about order updates, delays, and relevant promotions, ML-powered analytics that predict churn and surface upsell opportunities, and expansion to Instagram DM and Facebook Messenger on the same conversation engine. The goal: keep the automation rate high while the query catalog expands as Eona.ae grows.
“Working with EdgeFirm was a major improvement for Eona.ae. They developed a highly responsive WhatsApp chatbot that now powers our entire customer service and delivery communication. From automating FAQs and order updates to creating a smooth handover system for live support, their team delivered a solution that feels natural and intuitive to our users. Their ability to integrate real-time delivery data and build performance dashboards helped us scale with confidence. We're extremely pleased with their work and highly recommend them.”
Leadership Team, Eona.ae
What changed after launch
Higher Efficiency
About 90% of customer queries are handled automatically by the bot, as reported by Eona.ae, cutting support costs by 70-80% with no additional agents hired.
Faster Response Times
Initial customer response time dropped to under 30 seconds for most interactions, as reported by Eona.ae.
Improved Customer Satisfaction
Customers appreciated real-time delivery updates and the convenience of support without needing to call or email.
Smooth Human Support
Handover cases were handled without conversation drops, maintaining a smooth, high-quality customer experience.
The Eona.ae WhatsApp chatbot redefined customer communication by merging automation and human empathy. Through thoughtful design and reliable integrations, it helped Eona.ae to deliver a superior customer service and delivery experience, setting a new standard for responsiveness and reliability in the UAE market.
More case studies
Get AI live in months, not years.
Book a 30-minute strategy call with the founders who will write your code.
