← All case studies
AI Automation · Case Study

How I Built an AI-Powered Sales Pipeline for My Own Business

12+ hrssaved per week
3 agentsworking autonomously
90%less manual research

If I'm going to tell hospitality operators that AI automation can transform their operations, I'd better be running my own business on it. Here's exactly how I use AI agents to handle lead generation, research, and outreach for On The Hill AI — and what it saves me every week.

The Problem

As a solo consultancy founder, I was spending roughly 15 hours a week on business development: manually searching for potential hospitality clients, researching each business, crafting personalised outreach emails, and verifying the information before sending. That's almost two full working days every week that weren't being spent on billable client work or building the product.

The irony wasn't lost on me. I was selling AI automation to reduce admin, while drowning in my own admin.

What I Built

I designed and built a multi-agent AI system using LangGraph and Python that handles the entire lead generation pipeline:

Lead Finder Agent: Searches for UK hospitality businesses matching my ideal client profile — multi-site operators, independent restaurants and hotel groups, catering companies. It pulls from multiple data sources, filters by location and size, and outputs a structured list of prospects with contact details.

Research & Pitch Generator Agent: Takes each lead and conducts deep research: their website, online reviews, social media presence, job postings, tech stack indicators. It then generates a personalised outreach email that references specific operational pain points relevant to that business — not generic templates.

Fact-Checker Agent: Reviews every generated email for accuracy. Checks that business details are correct, that claims are substantiated, and that the tone is appropriate. Flags anything that needs manual review before sending.

All three agents feed into a centralised AI Agent Dashboard that gives me a single view of the pipeline: leads found, emails drafted, emails verified, responses received.

The Results

  • Lead research time dropped from ~15 hours/week to less than 3 hours of review and approval
  • The system processes 60–80 leads per batch, compared to 15–20 I could manage manually
  • Email quality improved — the fact-checker catches fabricated details that manual drafting misses under time pressure
  • I can now focus my time on client delivery and building the product rather than business development admin

Why This Matters for You

This is the same approach I take with every client engagement. The tools and techniques are different depending on your business — you might need booking automation instead of lead generation, or supplier management instead of email outreach — but the methodology is identical: map the process, identify the repetitive work, build agents that handle it, and keep you in control of the decisions that matter.

I built this system because I needed it. I use it every day. And if it breaks, I'm the one who fixes it — which means I understand what reliability actually looks like in production, not just in a demo.

Want to see what automation could do for your business?

Book a free 30-minute call. We'll look at where your time is going and tell you honestly whether automation makes sense.

Book a free consultation