What I’m thinking about: I bought a $2,400 conference ticket before Serious AI has sent its first invoice…and the plan to turn it into a $2M client pipeline came together in under 48 hours.
As this lands, I’m walking into day two of the flagship conference for medical practices, carrying a printed package built for 14 specific organizations.
Serious AI Partners is only a few weeks old, aimed at mid-market companies (roughly $5M to $500M in annual revenue), and early interest has come from all over the map. A couple of mentors pushed back (fairly) on that breadth…as I’ll hit a supply constraint on who I can actually serve long before I run out of interested industries, so their advice was to pick two or three verticals now, focusing where the commercial opportunity is biggest and where I can be the absolute best.
Why healthcare goes first
Some of the reasons are specific to me. I trained as a physician (M.D., University of Illinois College of Medicine) and spent 2017 through 2019 in healthcare venture capital at FundRx (now MBX Capital), so this is a chance to dust off a degree I haven’t put to work in years, plus I’ll be able to re-establish relationships that can open doors. Also, one of the core skill sets in medicine is differential diagnosis (rule out what can kill before you treat what might heal), which is the same discipline as underwriting a land deal, or an AI build.
The rest is macro. Health care and social assistance added 680,500 jobs from March 2025 to March 2026, while total US payrolls changed little on net (per the BLS). As AI reshapes most of the economy, the sector caring for the sick, the elderly and the young keeps hiring, and some of the largest, steadiest cash-flowing companies in the world sit inside it.
Reading their exact thoughts
The day after that conversation on industry targeting, the mentors asked for a one-pager to pass along to a few potential healthcare clients. Claude Code built it, plus a dedicated healthcare page built around my medical background, aimed at provider organizations (pharma and clinical research organizations carry about the heaviest regulatory burden imaginable for any software, let alone AI, and HIPAA means extra guardrails on anything touching patient data regardless).
Claude’s first pass at the pain points sounded plausible, like staff pasting patient data into public AI models. But I wanted proof it was still a live problem, phrased the way these operators actually describe current issues, so I had Claude go deep on late-2025 and 2026 data.
It came back with receipts. In a January MGMA poll, 56% of practices had neither a formal AI policy nor one in development, and the ones using AI without a policy were using it for documentation and prior authorizations. The first draft had the right instinct and no receipt, and the data turned a guess into a number.
That’s the Ask Method, the classic internet marketing strategy I learned from Ryan Levesque over 10 years ago…when prospects read your page, it should read like their own thoughts (…because the marketing copy is BUILT from their own words). What used to take weeks of surveys or hours of manual processing of Amazon reviews (speaking from experience) now takes a simple directive to an LLM.
Much of what surfaced was billing and payer friction, which rings true for me as a patient too (we’re still disputing a claim from our son’s hospital stay when he was born, and the back-and-forth with our insurance company continues to be a nightmare).
We also pointed our AEO (answer engine optimization, i.e. getting cited when someone asks ChatGPT or Claude a question) at those EXACT pain points instead of general searches like “Fractional Head of AI,” so the medical practice leaders asking about them start running into my name.
A week of progress, every day
I remind myself of this daily…at the pace these tools move, each day has to count like a week of progress.
So all of my healthcare lead-gen plans ran in parallel. One of those mentors suggested speaking at healthcare conferences and doing live builds on stage, so I handed Claude the transcript of the hot-seat session I had with my mentors and had it shortlist the conferences most likely to want an AI speaker.
Meanwhile, an Apollo.io search pulled mid-market medical organizations in Texas hiring for a Head of AI or similar role. All of it happened within ~24 hours of even considering healthcare as the primary vertical, and running these major research and processing tasks keeps getting cheaper (Claude Opus 5.5 shipped Sept 22, and Anthropic says it outpaces its larger Fable model on many benchmarks, at a lower price), so I can keep throwing more at the wall…while the quality of outputs keeps increasing (alongside my skill of wielding these tools).
The uncomfortable choice
The search for healthcare conferences turned up something almost too well-timed. The MGMA Annual Conference, the flagship event for medical practice leaders (physician groups, surgical groups, renal care providers and more), runs Sept 27-30 in San Antonio, a ~75-minute drive from me. No speaker slots left, but I couldn’t imagine a better room of potential clients.
My first reaction was resistance. I’ve been heads-down in my home office, and the whole thing kicks off on a Sunday, just a few days from when the option was put in front of me.
But I’ve started leaning on a simple rule, that the more uncomfortable choice is generally the right one. Recently, and especially since starting Serious AI Partners, at any decision point I ask, “What would the most aggressively determined version of myself do?” (…and that question now lives in the instruction file my Claude setup reads before every session, so it filters every call we make in this business).
Without question, the most determined version of myself would go to the conference, and play full-out. As Coach T (Trevor McGregor) puts it, “Your brain is a thermostat set at your comfort zone, and you need to turn up the heat to achieve anything in life.”
I thought the ticket was $100, until I realized that was an expired centennial promo. It was actually $2,400, and I bought it anyway, a few days out, before Serious AI has sent a SINGLE invoice.
14 targets before day one
Before paying, I had Claude pressure-test the whole idea. I’ve been out of the healthcare scene for several years, and I had no idea how legit this conference was or even what my game plan should be for attending. There’s no full attendee list, so Claude researched every speaker instead, tiering each by their public footprint (organization, website, Google reviews) and flagging where AI wasn’t anywhere near optimized (or in some cases, not in use at all).
More than 100 candidates came down to a tiered list of 14. It also screened all 171 vendors for direct competitors…including plenty of AI software firms (some with $30-50M Series A rounds behind them) that sell narrow tools into specific corners of a medical practice.
Our lane at Serious AI is the practice without a firm AI strategy yet, weighing those tools with no way to validate whether they’re worth the price, no strategy to get staff on board once they’re bought, or simply too busy to have anyone internally develop the know-how to intelligently implement AI within their organization. That gap is the whole point of a Fractional Head of AI.
For each of the 14 candidate organizations, Claude built a 4-page snapshot that “shows some skin” (i.e. how I’d use AI to fix what’s surfacing in their complaints, reviews and website gaps). Nothing overboard…a few minutes to read, and useful even if we never speak again.
(Nevertheless, my directive to Claude during the build-out of these snapshots was to ensure that the recipient would be blown away by the value, among other superlatives. I find subjective, simple prompts like this to work well for getting the desired output…obviously alongside my relentlessly detailed context-setting and pursuit of perfection.)
Each one gets printed, with a time-gated plan for when I’m most likely to cross paths with each target, alongside 100 healthcare-specific business cards, and (Claude’s suggestion) a handwritten note on top.
None of this strategy is new, it just runs MUCH faster now. Tim Ferriss has written that he used trade shows (SXSW, and the lounges at CES) to launch The 4-Hour Workweek onto the bestseller lists in 2007, the launch that made his career. As the story goes, he skipped the CES show floor entirely, sat in the blogger lounge, and pointed each blogger he met to the ~20 pages of the book most relevant to them.
I’m running the same play, with research capabilities he could only have dreamed of.
The floor under the bet
The plan is a ~$2M client pipeline from a $2,400 ticket. Whether it happens is anyone’s guess, but I couldn’t have built a better plan, and somewhere in the prep process the trepidation and discomfort turned into excitement…my eyes kept widening at how much bigger the opportunity was than I’d pictured.
Now it’s on me to execute (by the time you read this, I’ll already be a day in, and hopefully it’s going well). And the floor under the bet is high, because the absolute worst case is three days of healthcare operators describing their AI problems in their own exact words, which goes straight into how we build for them (and how our pages read to the next practice that finds us).
Our goal is to be the premier enterprise AI implementation agency in the world, and every decision gets weighed against that bar. This one was easy.
=====
If you run a physician group, surgery center or specialty practice where denials, prior authorizations and payer portals are eating the week, the work starts with diagnosing which of those is actually gating you, then building with verification where a licensed human still signs off. It’s how you become the practice whose AI work still holds up when a payer audits the claim.
Everything about how that works lives at seriousai.partners/healthcare, and if this reached you secondhand, please get in touch.