Karan Raj KR

Lead Intelligence Agent

Given a city and a category, it finds local businesses, scores the gaps in their online presence, re-checks the borderline ones, and drafts an email for each lead worth pitching.

For
KĀRYO’s door-to-door sales: it decides which doors are worth knocking on.
My part
Built alone for Agentathon 2026.
Built with
CrewAI, Groq, Python, OpenStreetMap

How it works

Four agents

  1. Researcher: finds businesses on OpenStreetMap, checks each website and domain age
  2. Scorer: rates the presence gap and the chance of converting, out of 20
  3. Manager: approves 16 and up, rejects 8 and below, re-researches the rest once
  4. Copywriter: one 100 to 140 word email per approved lead, naming its gap

Why

KĀRYO finds clients door to door. This agent decides which doors are worth knocking on, and what to say when one opens.

Decisions

A second look, not a coin toss

Borderline leads aren’t decided on one score. The manager asks a follow-up research question, the business is researched again and re-scored, and only then approved or dropped. Every decision is logged with its reasoning.

Explore Lead Intelligence Agent

More work