Bandbox is Kolkata's oldest and most trusted dry cleaning and laundry brand, founded in 1950 and operating for over 75 years. With 12+ outlets across Kolkata, Howrah, Siliguri, and Darjeeling, they command over 50% of Kolkata's organised laundry market. Despite their legacy, their customer operations — order bookings, pickup scheduling, status updates, and payment confirmations — were entirely manual, creating bottlenecks as order volumes scaled.
130+ hours
Staff hours saved per month
< 3 seconds (from 2–4 hours)
Average response time
+38% improvement
Order booking completion rate
84%
Customer query resolution without staff involvement
+22%
Repeat customer rate post-automation
Unlimited (was capped by staff headcount)
Peak hour query handling capacity
Bandbox was processing 300+ orders per day across 12 outlets — and every single customer interaction was handled manually by their staff. WhatsApp messages came in at all hours asking for pickup slots, order status, garment readiness confirmations, and pricing. The operations team was drowning.
Three problems were costing them time and money. First, a staff availability bottleneck: customer queries came in between 7 AM and 10 PM, and responses often took 2–4 hours during peak periods, leading to drop-offs before booking. Second, no order tracking visibility: customers had no way to check status without calling or messaging, requiring manual lookup across multiple outlet ledgers. Third, unstructured data: orders were tracked across physical registers and WhatsApp chats with no centralised record, so escalations, delays, and lost garments were handled reactively.
The management team was spending 130+ hours per month on WhatsApp coordination alone — time that should have been spent on operations and growth.
We built a fully automated WhatsApp AI agent using n8n, the WhatsApp Business Cloud API, and OpenAI GPT-4o — deployed on Bandbox's existing infrastructure with zero new hardware.
The system handles the entire customer conversation lifecycle autonomously. For order intake and slot booking, customers WhatsApp the Bandbox number and the AI agent identifies their request type — new order, status check, pricing query, or complaint — and routes accordingly. For new orders, it collects pickup address, garment count, and preferred time slot, then logs everything into a centralised Google Sheet accessible by all outlet managers in real time.
For real-time status updates, each order is assigned a unique ID at intake. When customers message asking for status, the agent queries the sheet, pulls the current stage (Pickup Scheduled, In Cleaning, Ready for Delivery, or Delivered), and responds within 3 seconds — 24 hours a day, 7 days a week.
The escalation engine handles edge cases: if the AI detects frustration signals, complaint keywords, or complex requests it cannot resolve, it immediately flags the conversation and pings the relevant outlet manager on a separate WhatsApp group with full context. No customer ever gets stuck in a dead-end loop.
For pricing and service queries, the agent draws from a full knowledge base of Bandbox's pricing, services, and outlet locations — so standard questions are answered instantly without involving staff.
The entire workflow runs on our self-hosted n8n instance on AWS EC2, integrated directly with Meta's Cloud API. We built fallback handlers for API downtime, message delivery failures, and edge-case conversation paths — ensuring the system degrades gracefully rather than going dark.
GPT-4o powered WhatsApp agent handles new order bookings end-to-end — collecting pickup details, slot preferences, and garment counts without any human involvement.
Customers get instant order status updates by simply messaging the Bandbox number. The system queries live data and responds in under 3 seconds, 24/7.
When the AI detects complaints, frustration, or complex edge cases, it immediately routes to the right outlet manager with full conversation context — no customer is ever left waiting.
Every order captured via WhatsApp is logged into a live Google Sheet accessible by all outlet managers — replacing scattered physical registers and WhatsApp threads.
The agent answers pricing, service, and outlet location queries instantly from a structured knowledge base — zero staff required for standard inquiries.
Built with API fallback handlers, message delivery retries, and graceful degradation paths so the system keeps running even during Meta API instability.
“Before this system, our team was glued to their phones all day just answering the same questions over and over. Now those hours are back, and our customers actually get faster responses than when we were doing it manually. Rishabh and his team understood our operations deeply before writing a single line of code.”
Aman Bahri
Director, Bandbox Group
Service area
This case study is part of our AI Automation practice. See how we approach similar projects, our process, and pricing.
Whether you have a clear brief or an idea on a napkin, we'd love to hear from you. Most projects start with a 30-minute call — no pressure, no sales pitch.
Related articles on this topic.
A precise, step-by-step tutorial for building a Shopify → Google Sheets sync with n8n. Covers webhook setup, field mapping, error handling, and the gotchas that every guide leaves out.
Read postMost agencies stop at launch. We think that's where the real work begins. Here's the exact n8n + Shopify + WhatsApp automation stack we built for Baby Forest India — with the production gotchas most tutorials skip.
Read postA step-by-step tutorial to build a production-ready 3-agent n8n pipeline: orchestrator, research agent (Brave Search), and writer agent (GPT-4o). Full node configs, system prompts, error handling, Docker Compose, and cost breakdown included.
Read post