Looking back to 2017—just a year prior to OpenAI developers launching the inaugural generative pre-trained transformer—artificial intelligence often conjured cinematic images like the T-1000 from Terminator 2. Predictions forecasting the imminent replacement of human radiologists by machines were already hitting feverish heights. During nearly every radiology conference back then, conversations routinely gravitated toward concepts like AI-driven workflows.

Nearly ten years down the road, those forecasts have expanded far beyond medical imaging to encompass virtually every sector of knowledge work. Moreover, the label "AI-driven" has now been compounded by "AI-powered," a fitting descriptor given how many enterprises construct their modern offerings upon foundation models engineered elsewhere.

A fresh analysis published jointly by SolComms and the HLTH conference, titled Healthtech Attention Economy, examines the public relations and media footprints of 70 early-stage American healthtech ventures spanning Series A through Series C funding. The study yields a stark conclusion: compelling messaging alone fails to guarantee media traction. Instead, a company’s LinkedIn follower count displayed the strongest correlation with press coverage. As the researchers pointed out, journalists gravitate toward leaders and enterprises already prominent within ongoing industry dialogues.

Out of 495 evaluated media pieces, a mere eight articulated tangible product outcomes delivered to a client. Half of that select group highlighted preliminary data stemming from Utah’s state pilot of Doctronic’s AI-assisted prescription tools. Meanwhile, a staggering 175 articles chronicled corporate milestones: funding rounds, strategic partnerships, product launches, corporate acquisitions, or initial deployments—collectively categorized as "events." Conspicuously absent from the spotlight is the unglamorous reality of spending two grueling years embedded side-by-side with clients, systematically fine-tuning workflows to maximize efficiency.

As an executive operating within the health technology sector, these findings prompt deeper, more foundational inquiries. With modern AI infrastructure lowering the technical barriers to entry for software creation, what truly separates one enterprise from another? How can investors and buyers accurately identify entities cultivating sustainable competitive moats when polished storytelling and slick product demonstrations are universally accessible?

When Technology Alone Is No Longer the Differentiator

In a recent exchange with Michael Burtov, chief innovation officer at Nova Southeastern University, the discussion turned to whether buzzwords like AI-powered retain any substantive meaning. Burtov remarked, "I think the default is AI. It’s just like saying I don’t use word processing in my business. Of course you should use word processing. It doesn’t matter what your business is."

Having founded his own artificial intelligence enterprise back in 2012, Burtov challenges the historical assumption that raw technology serves as the primary source of competitive advantage. Building a minimum viable product for a startup or replicating an established platform has become relatively frictionless. According to Burtov, the true business moat has rarely resided in the software itself; rather, it thrives within interpersonal relationships, customer networks, distribution channels, and the surrounding community.

Undoubtedly, exceptions persist where proprietary capabilities, distinct proprietary datasets, formal patents, rigorous clinical validation, or heavy regulatory hurdles erect formidable entry barriers. Yet, alongside the democratization of software development came a surge of superficial ventures. Burtov noted the proliferation of startup slop, where every pitch deck is meticulously polished and every product demonstration looks remarkably sophisticated.

Illustrating a widespread healthcare operational inefficiency, Burtov described scenarios where staff routinely resort to spreadsheets merely to bridge disconnected legacy systems. While such ad-hoc workarounds do not yet qualify as structured data, they represent crucial observational insights harvested at the operational grassroots. As he succinctly stated, observation constitutes the ultimate moat.

Two separate companies might theoretically leverage identical foundational models and public datasets, yet both can still lack the contextual, ground-level expertise required to solve a specific clinical pain point. Emma Silverman, a partner leading healthcare investments at venture capital firm TMV, noted that working directly with clients earns firms access to exclusive datasets absent from standard models. These insights dwell exclusively within day-to-day clinical workflows, reinforcing her belief that true corporate memory must be earned.

For years, professional services carried a negative connotation within venture capital circles. Traditional software-as-a-service (SaaS) models were prized because they scaled revenues without requiring a linear expansion in headcount. Today, however, enthusiasm has surged for forward-deployed engineers—skilled technologists who embed directly inside client environments to diagnose bottlenecks and engineer bespoke solutions. Silverman observed that very few venture-scale opportunities in modern healthcare operate entirely free from service-oriented mechanics.

Naturally, a delicate balance exists between uncovering universally solvable customer friction points and drifting into custom IT consulting. This hybrid discipline draws from technical, product, clinical, and analytical competencies. The overarching challenge for these multidisciplinary squads is to convert custom implementations into standardized capabilities that diminish the friction of onboarding subsequent clients. Within that operational translation lies the true foundation of sustainable competitive advantage.

Competing With the Incumbents

Tressa Springmann, chief information and digital officer at LifeBridge Health, constantly weighs the solutions offered by current vendors against anticipated future capabilities. She evaluates whether her primary platform partners currently offer a feature or plan to introduce it shortly, thereby maximizing returns on prior capital investments.

Such strategic alignment, however, does not preclude partnerships with agile startups, particularly when addressing critical, unresolved clinical challenges ignored by legacy enterprise partners. Springmann explained that if an issue remains stubbornly unsolved—perhaps driven by unique patient demographics or specialized clinical departments—they will engage a startup to learn and innovate rapidly.

For early-stage ventures, that temporal window spans roughly two years. It represents a strict timeline to prove efficacy and establish differentiation before massive electronic health record vendors or enterprise giants roll out native parity features. Once incumbents catch up, switching costs involve far more than simple feature comparisons; ongoing expenditures related to integration, enterprise governance, and long-term maintenance mean that matching an incumbent feature set is rarely sufficient for survival.

Reflecting on his past tenure directing radiology AI implementations at the University of Alabama at Birmingham—and currently serving as radiology chair at St. Jude Children’s Research Hospital—Andrew Smith shared a pivotal realization: a mere technological widget is never enough.

When an FDA-cleared diagnostic algorithm successfully identifies a critical radiological anomaly, it functions precisely as engineered. Yet, the downstream impact remains infinitely more important: Does that alert reach the appropriate physician promptly? Does it accelerate clinical intervention? Does it tangibly elevate patient health outcomes?

Smith emphasized that while a digital tool may perform flawlessly in controlled theoretical environments, execution and clinical implementation dictate ultimate success.

Returning to the insights of the SolComms report, publicly announcing a successful software deployment merely signals commercial sales capability, offering little visibility into actual clinical value generated. Mark Sendak, CEO of Vega Health and former leader at the Duke Institute for Health Innovation, noted a persistent disconnect between capital investments in healthcare technology and solutions genuinely optimized for clinical outcomes.

Even when a tool improves patient care, purchasing organizations frequently struggle to demonstrate a clear financial return. Conversely, technologies yielding verifiable cost reductions do not automatically enhance clinical metrics. Within the domain of clinical AI, isolating and attributing health improvements directly to a single piece of software remains exceedingly complex. This forces founders into a perennial dilemma: should companies be engineered around the clinical problems that matter most to patients, or around the administrative problems payers are guaranteed to fund?

When the Price Falls, Does the Market Shrink or Expand?

Having spent much of his professional career immersed in clinical documentation workflows, industry observers have watched production costs drop precipitously. Drawing from historical radiology data, generating a formal diagnostic report today often incurs a cost roughly 75% lower than the fully loaded operational expenses of the 1990s. Digital voice recognition previously eliminated workflows dependent on physical couriers, magnetic cassettes, and manual transcription pools, while modern ambient AI represents the next logical phase in that evolution.

Nevertheless, as clinical documentation expenses continue their downward trajectory, critical questions arise regarding the total addressable market size for firms operating within that domain.

Punit Soni, founder and chief executive officer of Suki, challenges the premise that falling unit costs erode market potential. Pointing to historical technological paradigms like the internet and digital memory, Soni noted that when individual task costs plummet, the activity becomes so ubiquitous that total market volume expands exponentially. Cheaper memory did not curtail memory consumption; rather, it unlocked entirely unprecedented use cases.

The lingering question, however, centers on how much of that newly generated economic value any individual startup can capture. A rapidly expanding market does not automatically equate to a defensible market position.

Concluding his dialogue with Soni, the discussion pivoted to primary performance metrics for privately held healthcare AI enterprises, stripping away vanity metrics like total capital raised or headline valuations.

Footprint. How many customers do you have? How many health systems do you have? Number one.

Soni designated customer footprint as the premier indicator of health, followed closely by Net Promoter Score and client satisfaction indices. His third core metric integrated gross margins, operating expenditure ratios relative to revenue, and overall growth velocity. Soni cautioned against evaluating growth in isolation, noting that rapid scaling achieved through unsustainable OpEx margins signals underlying structural fragility.

Drawing from firsthand industry trench work, additional evaluation criteria merit consideration: How prolonged and resource-intensive is a full organizational rollout, moving beyond limited pilot phases? Does implementation friction decline over successive deployments? What are the baseline customer retention rates and expansion revenue trajectories? Most importantly, is there rigorous, verifiable evidence that the software genuinely improves the underlying clinical labor it targets?

Quality metrics can carry equal weight to sheer quantitative expansion. Collectively, these multidimensional barometers indicate whether an enterprise is merely accumulating customer logos or forging scalable operational competencies that justify long-term industry viability.

A decade ago, the industry was content plastering trade-show booths and marketing banners with slogans about AI-driven workflows. Today, tangible evidence of real-world impact, hard-earned domain expertise, and deeply forged interpersonal relationships represent the true differentiators setting exceptional healthcare companies apart.

The underlying technology is more accessible than ever before. Everything else must still be earned.