How I Built an Automated Prospect Discovery System That Finds 250+ Leads on Autopilot

When you are building an AI services business, the biggest bottleneck is not the technology. It is finding the right businesses to talk to.

I spent weeks manually searching Google Maps for doctors, dentists, and law firms in South Florida. Copy a name, grab a phone number, check if I already had them, add to a spreadsheet. Repeat. It was brutal.

So I did what any developer would do — I automated the whole thing.

The Problem

I am building AI receptionist and review management tools for local businesses. My target market is professional services in South Florida: medical practices, dental offices, law firms, CPAs. There are thousands of them, and I needed a steady pipeline of qualified prospects to call.

Manual research gave me maybe 10-15 prospects per hour. I needed hundreds.

The Architecture

The system I built runs five times per day across four different campaigns. Each campaign targets a slightly different value proposition — AI receptionists, review management, general AI automation, and a straightforward efficiency pitch.

Here is how each run works:

1. Search: Query Google Maps for specific business types in specific cities — dentists in Boca Raton, personal injury attorneys in Fort Lauderdale, dermatologists in Palm Beach Gardens
2. Extract: Pull the business name, phone number, category, and location
3. Deduplicate: Check every prospect against the existing queue by phone number. This is the most important step
4. Enrich: Tag with campaign, category, and geography
5. Queue: Add to the outreach pipeline with all metadata

The whole thing runs on cron jobs. I wake up to a log of what was added overnight.

Deduplication Is Everything

Here is the thing nobody tells you about automated prospecting — the duplicate problem is worse than the discovery problem.

After the first week, roughly 40% of search results are businesses I already have in the queue. By week two, that number climbs to 60-70% in saturated geographies.

Without aggressive deduplication, you end up calling the same dentist three times from three different campaigns. That is not a good look.

I deduplicate on phone number, which handles most cases. Same business, different listing name? Caught by the phone number. Multiple locations of the same practice? Each gets its own entry since they have different numbers.

The Numbers

In about two weeks of running, the system has built a queue of over 250 qualified prospects. Here is the breakdown:

• Medical practices: ~80 (family medicine, internal medicine, dermatology, orthopedics)
• Dental offices: ~70 (general, pediatric, orthodontic, oral surgery)
• Law firms: ~65 (personal injury, family law, estate planning, immigration)
• Financial services: ~35 (CPAs, wealth management)

All in South Florida. All with verified phone numbers. All tagged by campaign and category.

What I Learned

Geography saturates fast. Fort Lauderdale law firms were tapped out within a week. You need to constantly expand your search radius and try new city/category combinations.

Category diversity matters. Starting with just doctors, dentists, and lawyers felt like enough. But adding CPAs, chiropractors, orthodontists, and dermatologists doubled the available prospects in the same geography.

Five runs per day is the sweet spot. More than that and you are mostly finding duplicates. Fewer and you are leaving prospects on the table. Each run adds 3-6 new businesses on average.

The real value is the queue, not the search. Anyone can search Google Maps. The competitive advantage is having a clean, deduplicated, categorized pipeline that is ready for outreach at any moment.

What Is Next

The discovery system feeds into an AI-powered outreach pipeline. Each prospect gets a personalized call from an AI agent that can have a natural conversation about their specific pain points — appointment scheduling for medical offices, review management for dental practices, after-hours coverage for law firms.

The whole stack is AI-native: AI finds the prospects, AI makes the calls, AI handles the conversations. My job is building and tuning the systems.

If you are building any kind of outbound sales operation, automate your prospecting first. It is the highest-leverage thing you can do. The technology is straightforward — the hard part is the deduplication logic and knowing when to expand your search parameters.

The best prospects are not the ones you find. They are the ones your system finds while you sleep.