As a seasoned architect of enterprise SaaS and a strategic voice in software design, Vijay Raina has spent years dissecting the mechanics of how technology truly integrates into the complex fabric of professional industries. With the current landscape of health tech shifting beneath our feet, he brings a grounded perspective to the often-frenzied discussion surrounding artificial intelligence. His expertise lies not just in the “how” of building software, but in the “why” of its long-term survival, particularly in sectors where reliability and workflow integration are far more valuable than mere novelty.
The following discussion explores the delicate balance between being an AI-native startup and a legacy software provider in a market that is rapidly maturing. We examine the current capital trends where a majority of digital health funding is tied to AI, the phenomenon of “point solution fatigue” among healthcare providers, and the historical parallels between this era and the “mobile-first” wave of the past decade. Through this lens, we uncover why the most durable companies are those that prioritize becoming a “system of record” over chasing the latest hype cycle.
With more than half of digital health funding now flowing toward AI-enabled ventures, how should founders navigate the intense pressure to be “AI-first” while maintaining long-term sustainability?
The pressure is palpable, and you can almost feel the desperation in boardrooms as founders scramble to slap an “AI-first” label on their landing pages. It is important to remember that in 2025, we saw a staggering 54% of all digital health funding pour into companies that claimed some form of AI enablement, which creates a distorted reality for those still building traditional infrastructure. My advice is to stop obsessing over whether you are “AI enough” and start focusing on whether you are “useful enough” to justify your place in a provider’s budget. You have to decide if you want to be the company making the individual screwdrivers or if you want to be the DeWalt of your niche, providing a comprehensive, powered ecosystem that workers rely on every single day. If you chase the “AI-first” tag without a deep understanding of the problem, you are essentially building a feature that a larger player will eventually swallow whole.
For established software companies that have already secured a place in clinical workflows, what are the most critical steps to bolster their defensibility against the new wave of AI-native startups?
Legacy software companies actually hold a much stronger hand than the current hype suggests, primarily because they already sit squarely within the daily, grueling processes of their clients. These organizations have spent years building deep workflow depth, securing proprietary data, and establishing the key integrations that make them a massive pain to remove or replace. If you are one of these companies, your mission is to use AI to harden those existing moats rather than trying to reinvent your entire identity. You should be looking at the data living inside your existing workflows—the kind of specialized information that a generic AI model cannot access—and using it to provide insights that an outsider simply cannot replicate. The goal is to elevate your status to a “system of record,” because once you are the source of truth for a practice, the “shiny new toy” startups will find it nearly impossible to dislodge you.
We are seeing a growing trend of “point solution fatigue” among healthcare providers. How does this impact the survival rate of specialized AI models that focus on niche tasks like claims submission?
The grace period for niche AI solutions is narrowing at an alarming rate as healthcare leaders realize they cannot manage fifty different dashboards for fifty different tasks. A specialized model for something like claims submission might look like a miracle on day one, but if that model doesn’t control the broader workflow or the underlying customer relationship, its value proposition begins to fade into the background. We are moving into an era where the market rewards businesses that show real usage and retention across a broad domain, not just a clever algorithm that solves one tiny slice of a problem. If you cannot prove a consistent, high-level ROI that justifies adding yet another vendor to a provider’s stack, your window of opportunity is closing. Point solution fatigue is the silent killer of AI startups that fail to transition into becoming defensible software companies with a wider footprint.
While the hype cycle suggests a total revolution, healthcare organizations are being quite selective. In what specific areas are you seeing the most successful real-world implementations right now?
The reality on the ground is far more pragmatic than the headlines, with healthcare leaders taking a very grounded view of where they deploy these tools. In 2025, we saw the number of organizations using AI jump from under half to more than two-thirds, but that growth was concentrated in very specific, “safe” areas like administrative work, transcription, and revenue cycle automation. Providers are saying “yes” to AI when it solves a clearly defined workflow need that carries manageable risk and offers a clear path to saving money. They have a very picky palate when it comes to advanced agentic tools that make clinical decisions, preferring instead the “Swiss Army knife” approach of AI that simply makes the paperwork disappear. Successful implementation right now isn’t about the most complex math; it’s about the most seamless integration into the boring, daily tasks that keep a clinic running.
You have compared the current AI wave to previous shifts like the “mobile-first” era. What lessons should we carry forward from the days of mobile-friendly EHRs to ensure we don’t repeat the same mistakes?
Looking back at the last decade, we saw companies like MacPractice and DrChrono gain huge momentum simply by being the “Apple-friendly” or “mobile-friendly” choice in a world of clunky legacy EHRs. They captured significant market share because they had a temporary technological differentiator, but eventually, the legacy players simply cannibalized those features and the playing field leveled out again. AI is following the exact same trajectory; it will eventually become ordinary and expected, much like having a mobile app is standard today. The lesson here is that a technological “edge” is not a permanent moat; it is merely a head start that you must use to build a real business. If your only value is that you use AI, you are destined to be a footnote once the established giants inevitably catch up and integrate those same capabilities into their existing platforms.
What is your forecast for the evolution of AI-native companies as they attempt to transition into “defensible software companies”?
I believe we are entering a “great consolidation” phase where the vast majority of AI-native companies will either be absorbed as features into larger stacks or quietly disappear as their niche value evaporates. To survive, these companies must pivot immediately from being “AI-first” to being “workflow-first,” proving they can manage the complexities of healthcare at scale, just as companies currently supporting 3,000 practices and 2 million patients have had to do. The winners won’t be the ones with the most sophisticated models, but the ones who successfully capture the “system of record” status by making themselves indispensable to the provider’s daily life. AI will stop being the headline and start being the engine under the hood, and by 2027, we will look back and realize that the most successful “AI companies” were actually just great software companies that knew how to use a new tool effectively.
