Skip to main content
Platform In Progress

LeadProc

AI-powered missed-call interception for home service contractors — captures and qualifies leads that would otherwise be lost to voicemail.

Role

Designer & Developer

Live site

View live site

Stack

Astro · Hono · Node.js · PostgreSQL · Telnyx · OpenAI · Stripe

LeadProc screenshot

Overview

Home service contractors — plumbers, electricians, HVAC techs, roofers — miss a significant share of incoming calls while they’re on a job. A missed call from a homeowner is a lead that almost always goes to the next contractor in the Google results.

LeadProc intercepts those calls. When someone doesn’t get through, LeadProc texts them immediately and opens an AI-handled SMS conversation — qualifying the lead, understanding the problem, collecting contact details, and flagging urgency. The contractor receives a structured summary before they’re even off their current job.

What Was Built

Telephony layer. Telnyx powers call forwarding and SMS delivery. When a call is missed, Telnyx fires a webhook to the application server, which sends the outbound SMS. Round-trip from missed call to first text: under 5 seconds.

AI conversation engine. OpenAI handles the SMS conversation. The agent is prompted with the contractor’s business context and qualification criteria. Conversations are stateful — context is maintained across multiple SMS exchanges within a session.

Lead qualification output. When a conversation ends (or times out), the system generates a structured summary: contact info, problem description, urgency level, confidence score. Delivered via push notification and stored in the contractor’s dashboard.

Multi-tenant architecture. Each contractor has their own phone number, business context, and qualification criteria. The data model separates tenants cleanly; AI context is configured per-tenant at onboarding.

Subscription billing. Contractors pay a monthly flat rate for a defined number of intercepted leads; overages bill at a per-lead rate via Stripe.

Technical Approach

Hono was chosen for its lightweight footprint and TypeScript-native design — for a webhook-heavy system where response latency matters, its minimal overhead is a real advantage. Telnyx over Twilio for pricing and webhook reliability. PostgreSQL for all relational data — the schema is deliberately simple for this use case.

Status

  • Core intercept-and-qualify flow: functional — missed call → SMS → AI conversation → lead summary
  • Contractor dashboard and billing layer: in active development
  • Live at leadproc.com

More screenshots

LeadProc screenshot 2

Need something similar built?

We work with businesses across industries — websites, platforms, and everything in between.