How We Built an AI WhatsApp Agent for a Jaipur Furniture Brand (Case Study)

This is the story of how Autozentic built Kia — an AI WhatsApp agent — for a mid-size furniture showroom in Jaipur, and how it changed how they handle sales inquiries. We're sharing the full case study: the problem, what we built, the technical approach, and the results.

The Problem: 4-Hour Response Times, Lost Leads

The client ran a furniture showroom in Jaipur with two sales staff managing all customer communication. On a typical day, they received 30–50 WhatsApp messages — a mix of price inquiries, stock availability questions, custom order requests, and delivery follow-ups.

The issues were clear: response times averaged 3–4 hours during peak periods. Customers who didn't hear back immediately sent the same inquiry to competitor showrooms and often bought from whoever responded first. Catalogues were being sent manually — the salesperson had to find the right product images, compile them, and send individually. Lead details were tracked in a shared spreadsheet that was constantly out of date.

The owner put it simply: "We're losing customers not because our products are worse, but because we're slower to respond."

What We Built: Kia, the WhatsApp AI Agent

Kia is an AI agent connected to the client's WhatsApp Business number. Here's what it does:

• Instant response — responds to every inbound WhatsApp message within seconds, regardless of time
• Intent detection — understands whether the customer is asking about price, availability, delivery, or wanting to speak to a human
• Dynamic catalogue sharing — pulls from the product database and sends relevant product images and pricing based on what the customer asked about
• Lead capture — collects customer name, location, product interest, and budget through a conversational flow
• CRM integration — every lead is automatically created in Furzentic with all captured details and tagged by product interest
• Follow-up sequences — if the customer doesn't respond to a quote within 24 hours, Kia follows up automatically with a soft reminder
• Human escalation — when a customer asks to speak to someone, wants to negotiate price, or makes a complaint, Kia flags the conversation and notifies the salesperson

The Technical Architecture

Here's how Kia was built:

• WhatsApp API access via an approved BSP (Business Solution Provider)
• n8n as the orchestration layer — receiving incoming messages, routing logic, triggering actions
• GPT-4 (via OpenAI API) as the language model powering Kia's understanding and response generation
• Furzentic CRM as the data layer — product catalogue, customer records, order history
• Custom knowledge base — the client's full product catalogue, FAQ, pricing, and delivery policies were indexed for Kia to query
• Memory layer — Kia maintains context within a conversation, so the customer doesn't need to repeat themselves

Total build time: 3 weeks from kickoff to go-live.

The Results (After 60 Days)

• Average response time: 2 minutes (down from 3–4 hours)
• Lead capture rate: +65% — more inquiries are now captured as CRM records with full details
• Salesperson time on WhatsApp: -70% — staff now focus on qualified leads that Kia has already engaged
• Conversion rate: +28% — faster response and consistent follow-up directly improved close rates
• Catalogue sharing time: 0 minutes — fully automated, 100% of relevant inquiries receive a catalogue within seconds

What We Learned

A few things stood out from this project:

• Data quality matters most. Kia is only as good as the product data it has access to. The first week was spent cleaning and structuring the client's catalogue properly — that work paid off significantly.
• Escalation paths are critical. An AI agent that can't hand off gracefully to a human frustrates customers. The escalation trigger design was one of the most important parts of the project.
• WhatsApp tone matters. Business communication on WhatsApp in Jaipur is informal and relationship-oriented. We tuned Kia's tone to match — conversational, warm, not robotic.
• Iteration is ongoing. We reviewed the first 500 conversations and found 3 new question types Kia wasn't handling well. Two weeks of prompt tuning fixed those gaps.

Want Something Like This for Your Business?

If you run a furniture business, retail store, or any business where WhatsApp is your primary sales channel — Kia's architecture can be adapted to your context. Talk to Autozentic about building a WhatsApp AI agent for your business.

Or if you're a furniture business specifically, explore Furzentic — our CRM built specifically for furniture retailers and manufacturers.