How I Used AI to Build a Pipeline of 30+ Acquisition Targets in Two Days

In my last month at GoHealth Urgent Care, I decided to test a new approach to deal origination. In about two days, I used AI to build a fully enriched sourcing list of 30+ independent urgent cares across a handful of highly specific regional markets. Then I used AI to customize every outreach message.

The response rates were incredible. Physician owners replied directly to my emails, specifically asking how I knew such granular details about their business, their history, and their local competitive positioning.

Historically, executing this kind of hyper-targeted sourcing required a $50,000 annual subscription to Definitive Healthcare combined with hours of manual research. Today, AI gets you 80% of the way there for almost nothing.

Here is exactly how I built this workflow, step by step.

We all know what manual sourcing looks like in healthcare M&A. You start by pulling massive CSV exports from Definitive Healthcare or PitchBook. Then you spend hours running manual Google searches to verify if these clinics are actually independent or quietly affiliated with a local health system.

After you figure out the corporate structure, you have to dig for contact info. You scour "About Us" pages or hunt through state registry filings trying to identify the actual physician owners or managing partners. Then you make cold calls with zero context or blast out generic templates that get ignored. It is a slow, expensive grind.

I replaced that entire manual process with a targeted AI workflow using tools like Perplexity, Claude, and ChatGPT. Here is the exact playbook.

Before opening any AI tools, you need strict parameters. For my GoHealth sprint, the mandate was narrow - independent, non-health-system urgent care clinics located in very specific geographies with a minimum location count.

The tighter your criteria, the better AI performs. Vague mandates produce vague results.

Instead of relying on static database exports, I used Perplexity to run deep contextual searches across the web. Perplexity is excellent at reading local news and clinic websites to verify independence.

Example Prompt (Perplexity):
"I am looking for independent urgent care operators in [target MSA]. Identify 15 clinics that have between 2 and 5 locations. Exclude any clinics owned by, branded with, or affiliated with [major health systems in the region or competitive PE backed platforms]. Provide the clinic name, website URL, and estimated number of locations."

Run this across each of your target markets. In my experience, Perplexity consistently surfaced operators that would have taken hours to find manually.

Once the AI pulled every potentially actionable target in the region, I then used it to enrich the list. AI identified the owners, extracted their contact information, and summarized their recent business moves.

Example Prompt (Claude or ChatGPT):
"Review this clinic website: [Insert URL]. Identify the primary physician owner or medical director. Extract their name and any available contact info. Then summarize their primary service offerings, mention any new clinic locations opened in the last two years based on their news section, and identify their main competitive advantage as stated on their site."

This is where the magic happens. You end up with a target profile that reads like you spent an hour researching each practice individually.

With the enriched data in hand, I had AI customize every single outreach message. The key is forcing the AI to reference the specific details from Step 3 so the email reads like it was written by a peer who did their homework.

Example Prompt:
"Write a concise, professional cold email to Dr. [Name], the owner of [Clinic Name]. Mention that I noticed their recent expansion into [Specific Neighborhood] and respect how they have maintained an independent footprint outside of the [Local Health System] network. Explain that I am exploring potential partnership or acquisition opportunities. Keep the tone peer-to-peer and direct. Avoid marketing buzzwords."

This is why physician owners were responding asking how I knew certain things about them. The outreach felt personal because it was built on real, specific research - just done in minutes instead of hours.

But here is the reality - you will not always have a direct email.

Physician owners of independent practices are not easy to reach. Their emails are often buried behind office managers, practice administrators, or generic info@ addresses. So you have to be resourceful.

What worked best for me was a combination approach. When I had a direct email, I used the personalized AI-drafted message above. When I did not, I went through the clinic's website inquiry form or called the front desk directly.

The key to getting through on a phone call or inquiry form is a short, direct message that signals credibility and intent without overexplaining. Something like:

"ATTN: Dr. [Name] - My name is Shawn Rothlis and I am reaching out on behalf of GoHealth Urgent Care. We are actively exploring opportunities to expand our footprint in [market] and would welcome the chance to discuss whether there is strategic alignment between our organizations."

That is it. No pitch deck. No three-paragraph overview of your platform. You are a peer reaching out with a specific reason. Physician owners respond to directness - they do not have time for anything else. And of course, my cell was always in my signature and of the 50% of founders that did respond, they called me (the other 50% responded in email).

From there, if the front desk asks what it is regarding, you keep it simple: "I am with GoHealth's corporate development team and wanted to connect with Dr. [Name] about a potential strategic opportunity." In my experience, that framing gets you transferred or gets your message passed along far more often than a cold voicemail.

Sometimes they say they will pass the message along and you have no idea if it will actually reach the doctor. That is fine. Use the moment to gather intel. Ask when Dr. [Name] is typically in the office so you can time a follow-up call. Ask if there is a direct email you can send more detail to - they will almost always say no, but it is worth asking. And ask who manages the practice's business operations or partnerships, because sometimes the path to the physician owner runs through a practice manager or office administrator who actually handles these conversations. Every call, even the ones that feel like dead ends, should leave you with more information than you had before you dialed.

The point is that AI handles the research and the personalized written outreach at scale. But when the direct channel is not available, you still need to pick up the phone and work the traditional route - just with significantly better context than you would have had before.

You never send AI-drafted emails blindly. Review each draft. Check that the AI correctly identified the owner and did not hallucinate a geographic expansion or a service line that does not exist. I spent about 60 seconds tweaking each email before hitting send.

The human review step is non-negotiable. AI will occasionally get details wrong, and one bad email can burn a relationship before it starts.

The results were immediate. In about two days I had a fully enriched pipeline of 30+ targets with owner names, contact info, and enough practice-level detail to make every outreach feel like I had been tracking them for months. The kind of work that would have taken weeks of manual research and potentially a $50K+ database subscription.

But here is what I want to be honest about. AI is not a magic button. It will hallucinate details. It will occasionally misidentify an owner or miss that a practice was quietly acquired by a health system or other competitor six months ago. The research still requires human judgment - you are just compressing the timeline from weeks to days and letting AI handle the mechanical lift.

The real advantage is not just speed. It is coverage. When you are sourcing manually, you naturally gravitate toward the obvious targets - the ones with the biggest web presence or the names you already know. AI does not have that bias. It surfaces operators you would have missed entirely because they do not show up in the usual places.

From what I have seen, the teams that figure out how to layer AI into their sourcing workflows are going to build pipelines that are wider, deeper, and more personalized than anything a traditional database export can produce. And they will do it at a fraction of the cost.

Next week I am breaking down how AI can restructure and accelerate the comp table workflow in healthcare M&A - turning days of formatting into minutes while keeping the judgment calls where they belong.

If you found this useful, share Healthcare M&AI with a colleague who runs corp dev or deal sourcing. The more people rethinking these workflows, the better.

-Shawn