At Daalchini, we power over 2,00,000 vending machine orders every single day across India. In this blog we are going to share that how we Solved 2,000 Daily Customer Tickets with Just 5 People and AI Magic
Now here’s the math: even if just 1% of those orders run into issues (payment stuck, product not dispensed, or order mismatch), that’s 2,000 customer tickets to be solved daily. Traditionally, this would need a team of 15–20 full-time agents, assuming each agent handles 100–150 queries a day.
We do it with just 5 support agents and 1 team lead! And here’s how we made it happen, not with brute force, but with intelligent tech, smart AI, and an automation bias.
Step 1: The Support Bot That Never Sleeps
We began by building an AI-powered chatbot that talks to customers across channels, including websites and apps, including WhatsApp. This bot isn’t just a glorified FAQ. It understands context, asks the right questions, and automatically tags the nature of the complaint.
Step 2: AI That Reads Like a Human
Most customers upload screenshots of payments or photos of the vending machine screen, so we trained OCR (Optical Character Recognition) tools to extract transaction IDs, machine IDs, and timestamps, all from those images.
Better yet, we now use AI-based image interpretation to identify:
- Was the snack stuck in the coil?
- Was the QR code not scanned properly?
- Was the machine’s screen showing an error?
From image → to issue → to suggested resolution, ALL in seconds.
Step 3: Auto-Resolutions That Just Work
Once the system identifies the issue, it automatically triggers refunds, credits the Daalchini wallet, or sends a re-attempt command to the machine.
This workflow now auto-resolves 37% of all tickets, without a human ever stepping in.
Step 4: Empowering Customers & Clients With DIY Tools
Not all users want to chat. Some just want to upload their payment screenshot and be done.
So we built a Self-Care Portal (a DIY platform) for:
- Customers who faced a failed order
- Franchise partners who manage machines
- Corporate clients who serve 100s of employees via Daalchini
Here, the AI scans the uploaded screenshot, identifies the issue, and resolves it autonomously, without waiting for support intervention.
Today, 10–14% of daily tickets are resolved via this DIY route, which is fast, trusted, and increasingly popular!
70% Fewer Support Agents, 2x Better Customer Experience
In total, nearly 50% of all our support tickets now get resolved without human help. And the best part is that the remaining 50% are now faster, smoother, and better handled because our human team is focused only on the complex, edge cases.
The Vision Ahead
This is just the start. We’re now exploring:
- Predicting complaints before they occur using machine learning
- Real-time machine health diagnostics via sensors and AI
- A voice-AI bot for vernacular support in multiple Indian languages
Afterword
Customer support doesn’t scale linearly with orders, but intelligent systems do. And at Daalchini, that’s how we’re building for the next million users, with AI, empathy, and speed.