OpenOutreach: an autonomous B2B lead generation system that discovers and qualifies prospects using Bayesian ML and LLMs

What it solves

OpenOutreach automates the B2B lead generation process, removing the need for pre-existing contact lists. It solves the problem of manually searching for, qualifying, and contacting potential customers by using AI to discover leads on LinkedIn and reach out via email or LinkedIn messages.

How it works

The system uses a multi-step AI pipeline:

1. Discovery: An LLM generates LinkedIn search queries based on a product description and target market.
2. Qualification: A Bayesian ML model (Gaussian Process Regressor) analyzes profile embeddings to identify ideal customers. It uses an explore/exploit strategy to refine its selection over time, while an LLM classifies the profiles.
3. Routing: Qualified leads are routed to either an email channel (if a work email is resolved via API) or a LinkedIn connection channel.
4. Outreach: AI agents generate personalized messages and manage multi-turn follow-up conversations.
5. Automation: Playwright with stealth plugins mimics human behavior to avoid detection, and a built-in CRM tracks the state of every lead.

Who it’s for

It is designed for founders, sales teams, and agencies who want to automate their outreach without relying on expensive subscription services or risking account bans through undetectable automation.

Highlights

• Autonomous Lead Discovery: Finds leads automatically from a simple product description.
• Bayesian Active Learning: Uses a Gaussian Process model to get smarter at picking leads with every decision.
• Email-First Approach: Prioritizes high-volume email outreach over LinkedIn connection requests.
• Self-Hosted CRM: Includes a built-in Django-based CRM for full data ownership.
• Stealth Automation: Mimics real user behavior to reduce the risk of account bans.