Project 29 / 33

LeadFlow

AI lead qualification and routing

An AI lead-qualification demo where the model classifies inbound leads and deterministic scoring ranks them.

LeadFlow — AI & Automation project cover

04 desktop / 02 mobile

Desktop

LeadFlow — desktop interface 01
Desktop 01
LeadFlow — desktop interface 02
Desktop 02
LeadFlow — desktop interface 03
Desktop 03
LeadFlow — desktop interface 04
Desktop 04

Mobile

LeadFlow — mobile interface 01
Mobile 01
LeadFlow — mobile interface 02
Mobile 02

Overview

LeadFlow is an AI lead qualification and routing demonstration. The model returns discrete categories from a closed enum, and a deterministic scoring rubric maps those categories to a 0–100 score. Relay is the fictional CRM product associated with the demonstration.

Approach

The AI classifies; code scores.

Highlights

07

  • Closed-enum classification
  • Deterministic scoring rubric: company fit 30, industry fit 25, intent clarity 25, budget signal 20
  • Spam is a hard override to zero
  • Unknown category values fail safely rather than producing invalid scores
  • Routing: 70 or above qualified / automated response; 30–69 human review; below 30 archived
  • Groq primary with Gemini fallback
  • Supabase, Resend, HMAC-cookie admin access, Zod validation and rate limiting

Limitations

02

  • Relay is a fictional CRM portfolio product and functional demonstration, not a commercial deployment.
  • An unused lead-enrichment stub remains in the codebase.