🚀 I Built an AI Team That Finds Leads, Decision Makers & Writes Outreach While I Sleep

Lead generation is one of those tasks everyone knows they should automate… but most teams still do it manually.

Every week the process looks something like this:

🔎 Search LinkedIn for companies

🌐 Visit dozens of websites

👤 Find the CTO, VP, or Director

📧 Hunt for contact information

✍️ Write personalized outreach emails

⏳ Lose half a day doing repetitive work

What if all of that happened automatically?

What if you woke up Monday morning to find:

✅ A list of companies matching your ICP

✅ Decision makers already identified

✅ Company research already completed

✅ Lead qualification scores calculated

✅ Personalized outreach drafts ready to send

That’s exactly what I built using Hermes Agent.

Full Video Walkthrough:

🤖 The Goal

I wanted a system where I could provide a simple campaign brief and have a team of AI agents handle the entire lead generation workflow.

Something like:

“Find SaaS startups with 10-200 employees that build AI developer tools. Identify decision makers and prepare personalized outreach.”

Instead of manually managing every step, a multi-agent workflow handles the process from start to finish.

🏗️ Architecture Overview

The pipeline consists of six specialized AI agents:

🎯 Orchestrator Agent

Acts like a sales manager.

Responsibilities:

• Creates execution plans


• Creates tasks


• Assigns work to specialist agents


• Tracks progress


• Manages dependencies


• Controls workflow phases

🔍 Prospector Agent

Finds companies that match your ICP.

Input:

• Keywords


• Industry


• Company size


• Target market

Output:

• Qualified company list

Example keywords:

• AI Developer Tools


• LLM Infrastructure


• DevOps Automation


• AI Engineering Platforms

🌐 Scraper Agent

Researches every company discovered during prospecting.

Collects:

• Products


• Services


• Company descriptions


• Locations


• Social profiles


• Technology signals

All data gets enriched and stored automatically.

👥 Contact Finder Agent

Identifies the right people inside each company.

Targets:

• CTOs


• VPs


• Directors


• Founders

Then gathers available contact information from multiple sources.

✉️ Outreach Agent

Generates personalized outreach emails.

Instead of generic templates, it references:

• Company initiatives


• Product offerings


• Technology stack


• Industry positioning

Result:

Much more relevant outreach messages.

📊 Analyst Agent

Scores every lead against the Ideal Customer Profile (ICP).

Evaluation criteria:

• Company size


• Industry fit


• Product relevance


• Buying potential


• Strategic alignment

Each company receives a qualification score between 0 and 1.

This helps prioritize outreach efforts.

🧠 Why Multi-Agent Systems Work So Well

Most people try to build lead generation using a single AI agent.

The problem?

One agent becomes responsible for:

• Research


• Scraping


• Qualification


• Personalization


• Coordination

That quickly becomes messy.

Instead, I use specialist agents.

Each agent focuses on one responsibility only.

Benefits:

✅ Better task quality

✅ Easier debugging

✅ Better scalability

✅ Parallel execution

✅ Cleaner workflows

📋 Workflow State Machine

The workflow runs in phases:

Campaign Brief

      │

      ▼

 Prospecting

      │

      ▼

 Research & Enrichment

      │

      ▼

 Contact Discovery

      │

      ▼

 Outreach Generation

      │

      ▼

 Lead Qualification

      │

      ▼

 Campaign Report

The orchestrator only unlocks the next phase once the previous phase is completed successfully.

This prevents bad downstream data from contaminating later stages.

🗂️ Hermes KANBAN Board = Shared Agent Memory

One of my favorite parts of Hermes Agent is its Kanban workflow system.

The Kanban board acts as a shared coordination layer between agents.

Every agent can:

📖 Read task status

✍️ Update progress

🔄 Create follow-up tasks

🚦 Track dependencies

The orchestrator uses the board to understand:

• What is complete


• What is blocked


• What should happen next

This creates a surprisingly robust autonomous workflow.

⚡ Running a Campaign

To launch a campaign I simply provide:

Product

What I’m selling

ICP

Who I want to target

Discovery Keywords

Where prospecting should begin

Goal

What outcome I want

Example:

Run a full B2B lead generation campaign for ShipMe Agent, an AI agent for QA and DevOps automation. 

ICP: AI SaaS startups, 10-200 employees. 

Keywords: AI Developer Tools, LLM DevOps Automation. 

Goal: 10 ranked qualified leads with decision-maker contacts and personalized outreach emails drafted

The orchestrator takes over from there.

📈 End Result

At the end of a run I receive all information organized in CSV files:

Companies

• ICP matched companies


• Enriched company data

Contacts

• Decision makers


• Contact information

Outreach

• Personalized email drafts

Qualification

• Lead scoring


• Prioritized opportunities

Report

• Campaign summary


• Workflow results

All generated automatically from a single task.

🔥 Real Value

The biggest win isn’t saving a few minutes.

It’s eliminating repetitive work entirely.

Instead of spending hours every week:

❌ Searching

❌ Researching

❌ Copy-pasting

❌ Writing first drafts

I can focus on:

✅ Sales conversations

✅ Closing deals

✅ Improving campaigns

✅ Building relationships

The AI team handles the operational work.

🛠️ Resources

🔗 Agents SOUL.md files & Custom Hermes Skills: https://github.com/vivekshetye/hermes-lead-generation-pipeline

💬 What Would You Automate?

If you had a team of AI agents working for you 24/7, what business workflow would you automate first?

Lead generation?

Customer support?

Market research?

Content creation?

I’d love to hear what you’re building.