LeadRadar AI

An AI-powered lead intelligence platform that replaces keyword monitoring with conversational intent discovery and semantic matching on Reddit.

95%Noise reduction vs keyword matching

<2sLead classification latency

0Manual keyword configurations

Overview

LeadRadar AI is an AI-powered Lead Intelligence Platform that continuously monitors Reddit and identifies high-intent opportunities for freelancers and small agencies. Unlike traditional keyword monitoring tools, LeadRadar AI's core differentiator is AI-powered Intent Discovery. Preferences configuration is performed entirely through a natural AI conversation on the dashboard, translating user responses into an internal Search Intent Profile and evaluating opportunities based on semantic relevance, problem alignment, and buying signals.

Technologies used

FastAPINext.js 16React 19Tailwind CSS v4PostgreSQLpgvectorCeleryRedisZustand

The challenge

Understanding the friction.

Traditional lead finding tools rely on exact keyword matching (e.g., F5Bot), which yields high noise, irrelevant alerts (tutorials, homework, job listings), requires complex manual include/exclude rule configurations, and provides no contextual understanding of buying intent, project value, or urgency.

““Moving from brittle keyword searches to conversational intent discovery completely shifts how freelancers find their next client.””

The solution

Engineered with focus.

Engineered a dual-engine architecture: an AI Conversation Engine to conduct dynamic onboarding chats (generating Search Intent Profiles) and a Search Intent Engine for semantic post matching. Developed an asynchronous pipeline with Reddit post collectors, Celery workers on Redis queues, and pgvector embeddings for semantic context search. Built a real-time lead qualification flow scoring posts from 0-100 on buying intent, value, and urgency.

Conversational AI OnboardingReplaces multi-field filters with a 6-question dynamic setup chat to extract business capabilities, target customers, and exclusions.

Explainable Match ReasoningCalculates 0-100 scores and shows bulleted explanations ("Why this lead?") detailing persona matches, intent strength, and budget metrics.

Outreach Reply GeneratorGenerates context-aware, personalized outreach messages tailored to the Reddit post's problem and the user's specific services.

Asynchronous QueueingUtilizes Celery, Redis, and a Reddit collector to poll, embed, match, and deliver instant Telegram or email alerts.

Results

Real world outcome.

Built a production-ready lead intelligence system featuring a conversational dashboard, structured match explanations ("Why this lead?"), AI outreach reply draft generator, pipeline CRM tracking, and automated Telegram alerts. Ensured precise lead filtering with a SQLite-compatible safe-vector design for local testing and scalable PostgreSQL deployments.

95%Noise reduction vs keyword matching

<2sLead classification latency

0Manual keyword configurations

Building something similar?

Book a free 30-minute call. I'll scope your project, identify risks, and tell you honestly if I'm the right fit.

Book a free call See all work →