# Scout: AI-Powered B2B Lead Generation Platform

> An AI agent that automates the entire B2B sales pipeline, from discovering companies and enriching contacts to qualifying leads and drafting personalized cold outreach.

URL: https://larch.ee/work/scout/
Published: 2025-03-01
Studio: larch (https://larch.ee/) · hello@larch.ee

## The Problem

B2B sales teams spend most of their time on repetitive grunt work: manually searching for companies, hunting down decision-maker contacts, qualifying leads against their ideal customer profile, and writing cold emails. It's tedious, inconsistent, and doesn't scale.

## Our Solution

Scout is an AI-powered platform that automates the entire lead generation pipeline. Point it at your ideal customer profile, and it discovers companies, enriches contacts, scores leads, and drafts personalized outreach, all autonomously.

## How It Works

### 1. Discovery

The AI agent searches across configurable sources: Product Hunt, Y Combinator, Crunchbase, GitHub, AngelList, Reddit communities, and more. It uses smart source optimization, deprioritizing low-performing channels and doubling down on what works.

### 2. Enrichment

For each discovered company, the system scrapes the web to find founders, CEOs, and key decision-makers. It extracts LinkedIn profiles, email addresses, and company details using Playwright with stealth mode and the Serper API.

### 3. Qualification

Every lead is scored (1–5) against the user's Ideal Customer Profile, considering industry, company stage, team size, budget, tech stack, and founder type. The AI explains its reasoning for full transparency.

### 4. Graph Expansion

Top-scoring leads seed a second wave of discovery. The system finds competitors, companies from the same accelerator cohorts, and related players in niche verticals, expanding the pipeline with high-quality lookalikes.

### 5. Outreach

The AI drafts personalized cold emails using the user's knowledge base: uploaded case studies, pricing docs, and company profiles. Emails are tailored to each lead's specific situation, not generic templates.

## Key Features

- **Kanban Pipeline**: Drag-and-drop lead management (Qualified → Contacted → Replied → Meeting → Won/Lost)
- **Knowledge Base**: Upload company docs, case studies, and testimonials to personalize all AI outputs
- **Email Sequences**: Multi-step follow-up campaigns with SMTP integration
- **Analytics Dashboard**: Pipeline health, conversion funnel, source performance, and outreach metrics
- **Activity Log**: Complete audit trail of every agent action
- **7-Step Setup Wizard**: Guided onboarding from ICP definition to first lead batch

## Technical Highlights

- **Next.js 15** with App Router and React 19
- **AI Agent Architecture**: Modular pipeline with discrete steps (discover → enrich → qualify → expand → draft)
- **Multi-cycle Automation**: Runs 10+ cycles autonomously to hit lead targets
- **Playwright + Stealth**: Web scraping that bypasses bot detection
- **SQLite**: Zero-config, file-based database with WAL mode for concurrency
- **OpenClaw Gateway**: Abstracts Claude/GPT models for flexible AI provider switching
- **Real-time Progress**: Live UI updates during agent runs

## Results

Scout runs the SDR pipeline on its own: it finds companies, enriches contacts, qualifies leads and drafts the outreach. Sales teams define their ICP once, upload their knowledge base, and let the agent build a qualified pipeline with personalized outreach ready to send.