Find buyers before they find your competitors.

LeadEcho monitors Reddit, Twitter/X, LinkedIn, Hacker News, Dev.to, Lobsters, and Indie Hackers for people actively describing the exact problem your product solves — then scores them by conversion probability, drafts AI replies from your knowledge base, and lets you post them with one click via the Chrome extension.

→ Live demo · Product Hunt · Docs

Features

• Semantic pain-point matching — Voyage AI embeddings match posts by meaning, not keyword strings
• 4-stage AI scoring — spam filter → semantic match → LLM intent classification → lead qualification
• Smart tiered inbox — Leads Ready / Worth Watching / Informational, auto-ranked
• Chrome extension + side panel — passive signal collection, reply queue with one-click posting
• Human-mimicry typing — Gaussian keystroke timing so replies look natural
• UTM attribution — unique short links per reply, track clicks → signups → revenue
• Knowledge base RAG — AI replies grounded in your own docs, not hallucinated
• Webhook notifications — Slack, Discord, email alerts for high-intent mentions
• BYOK — bring your own OpenAI, Anthropic, or compatible API key
• Authenticated crawling — Reddit, Twitter, LinkedIn via browser session cookies

Self-host in 5 minutes

Prerequisites

• Docker + Docker Compose
• An OpenAI or Anthropic API key (for AI scoring)

1. Clone and configure

git clone https://github.com/your-org/leadecho
cd leadecho
cp .env.example .env

Open .env and set at minimum:

JWT_SECRET=<random 32+ char string>
OPENAI_API_KEY=sk-...          # or ANTHROPIC_API_KEY
RESEND_API_KEY=re_...          # optional — for welcome emails

2. Start services

# Start Postgres + Redis
docker compose up -d

# Run database migrations
make migrate-up

# Build and start API + dashboard
make up

3. Register and onboard

1. Go to http://localhost:3100 and click Get Started
2. Register with email/password
3. Follow the 4-step onboarding wizard to set up pain-point profiles and keywords
4. The monitor starts scanning Reddit + HN automatically every 5 minutes

Platform setup

Reddit & Hacker News (no setup)

Work out of the box using public APIs. No account needed.

Dev.to, Lobsters, Indie Hackers (no setup)

Also public — no authentication required.

Twitter/X and LinkedIn (authenticated)

These require the Pinchtab browser sidecar:

# Start with browser sidecar (needs PINCHTAB_TOKEN in .env)
docker compose --profile browser up -d

Then in the dashboard, go to Settings → Browser Sessions and paste your:

• Twitter/X: session cookie (auth_token from browser devtools)
• LinkedIn: li_at cookie

LinkedIn stealth mode (Pro)

For less aggressive LinkedIn detection, use the Camoufox sidecar:

CAMOUFOX_URL=http://localhost:9868 docker compose --profile browser-pro up -d

Chrome Extension

Install from the Chrome Web Store (link coming) or load unpacked:

cd extension
pnpm install
pnpm build
# Load the dist/ folder in chrome://extensions → "Load unpacked"

Then in the extension Settings tab:

• Backend URL: http://localhost:8090 (or your hosted URL)
• Extension Key: generate one in the dashboard under Settings → Extension Token

Architecture

┌─────────────────────────────────────────────────┐
│  dashboard/ (React, Vite, TanStack Router)      │
│  landing/   (static HTML marketing site)        │
│  extension/ (WXT + MV3 Chrome extension)        │
└────────────────┬────────────────────────────────┘
                 │ HTTP
┌────────────────▼────────────────────────────────┐
│  backend/ (Go, Chi, pgx/v5)                     │
│  ├─ api/          HTTP handlers + middleware     │
│  ├─ auth/         JWT + Google OAuth             │
│  ├─ monitor/      9 platform crawlers           │
│  │   └─ scorer/   4-stage AI scoring pipeline   │
│  ├─ browser/      Pinchtab + Camoufox clients   │
│  ├─ embedding/    Voyage AI client               │
│  └─ database/     sqlc-generated queries         │
└────────────────┬────────────────────────────────┘
                 │
   ┌─────────────┴──────────────┐
   │ PostgreSQL (pgvector)      │
   │ Redis (session cache)      │
   └────────────────────────────┘

Monitor dispatch (every 5 minutes)

4-stage scoring pipeline

1. Rules filter — minimum length, spam patterns
2. Semantic match — Voyage AI cosine similarity against pain-point profiles (threshold: 0.40)
3. Intent classification — LLM labels: buy_signal, recommendation_ask, complaint, comparison, general
4. Lead qualification — auto-creates lead for score ≥ 7.0

Environment variables

Development

# Install all dependencies
make install

# Start everything (DB + API + dashboard)
make up

# Watch logs
make logs

# Run migrations
make migrate-up

# Generate sqlc types after SQL changes
make sqlc

# Type check frontend
make web-check

# Run Go tests
make api-test

# Format code
make fmt

Tech stack

License

MIT — see LICENSE

Contributing

Issues and PRs welcome. Please open an issue first for substantial changes.