A Saudi software company · 2026
Turning a coin-flip WhatsApp inbox into a tracked sales pipeline
A custom AI sales chatbot wired into a CRM, built on a 48-hour MVP sprint model, so inbound WhatsApp leads get captured, qualified, and followed up on instead of depending on whoever happens to see the message first.
Role
AI Engineer (Contract) — chatbot architecture, CRM pipeline, lead capture
Year
2026
Stack
Claude API, Next.js, CRM/lead pipeline, WhatsApp Business API
Overview
A growing software company based in Saudi Arabia was running sales almost entirely through WhatsApp, with inbound questions about pricing and availability landing in one inbox and getting answered whenever someone happened to be free. Nothing tracked who'd been contacted, what they asked, or whether anyone had followed up.
The ask was a sales chatbot, but the real problem underneath it was structural: every lead was a coin flip, and there was no record of the conversation once it ended.
Approach
Built a custom AI agent connected directly to the business's WhatsApp number, trained on their actual product and pricing information rather than a generic script, so it could answer real questions instead of deflecting to a human by default.
Wired the agent into a CRM pipeline built around how the business actually qualifies and follows up on a lead, so every conversation and contact gets logged automatically instead of living in one person's chat history.
Delivered on a 48-hour MVP sprint model: scope and deadline agreed up front, a working MVP shipped in 48 hours, with refinement running alongside the next sprint rather than blocking it.
Result
The business now has a system catching and logging every inbound WhatsApp lead automatically, instead of relying on whoever happens to be holding the phone.
This is an active, ongoing engagement — formal before/after metrics (response time, handoff rate, leads captured) are being tracked as the build continues, and will be added here once there's a confirmed number to publish.