O.C. Manjos Lead Qualifier

Inspiration

I run O.C. Manjos, an electrical merchant in Abuja. We have a WhatsApp AI handling sales enquiries — but high-value leads often go cold when a customer stops responding after a single message.

What it does

The skill pulls recent enquiries from Supabase, matches free-text product mentions against a real price sheet, and scores each one against category, price, recency, and disqualifier rules. Qualifying leads are handed to CALL-E's plan_call and run_call to confirm interest and delivery preference by phone.

How I built it

Node.js pipeline reading from Supabase and an Excel price sheet, with keyword/variant extraction for messy real customer text. Every threshold in the scoring rule is derived from real product pricing, not guesses.

Challenges I ran into

Real customer data is messy — most enquiries aren't clean product names, they're questions, confirmations, or payment chat. I had to build a keyword filter that correctly skips noise instead of misfiring on it.

Accomplishments that I'm proud of

Placed a real, successful end-to-end test call: the agent confirmed interest and captured a delivery preference from a live conversation.

What I learned

How to work with a brand-new, still-evolving platform — including hitting and working around real service issues (billing outages, region support changes, timeouts) during development.

What's next for O.C. Manjos Lead Qualifier

Wiring the qualification result to trigger CALL-E automatically, and connecting real phone numbers once WhatsApp Business API is integrated with Supabase.

Built With

• call-e
• javascript
• node.js
• supabase
• xlsx