Adoption of AI in healthcare continues to rise. Two in three physicians now use some form of health AI, up 78% from 2023, with healthcare organizations investing billions in AI initiatives across clinical, operational and administrative workflows.
Yet beneath the surface lies a concerning trend. According to a recent MIT report, enterprises are investing between $30 and $40 billion in generative AI, but over 95% are not seeing any returns from these investments.
Why many AI initiatives fail in healthcare
This begs the question: if AI holds such transformative potential for healthcare, why are most organizations failing to capture its value?
Broadly, when AI projects fail, we see five predictable patterns emerging:
- Low-impact workflow selection happens when organizations pick tasks that look easy to automate but don't matter much to the people using them. You can build technology that works flawlessly, but if it's not solving a real problem that end users care about, they won't use it.
- End user friction kills adoption when AI forces clinicians or staff into yet another system with its own login and interface. Healthcare workers already juggle too many applications and alerts. They need AI built into their existing tools, not one more thing competing for their attention.
- Missing integration pieces compromise projects built in isolation from core data systems. Your AI might test perfectly, but if it can't talk to the EHR, practice management, billing platform or other critical systems, it's an expensive experiment that will not be useful in the real world.
- Misalignment with the organizational mission starves projects of the executive backing they need to succeed. Cost reduction, better access, improved quality – whatever the strategic goal, AI initiatives must tie directly to it. If your C-suite can't explain how a project moves the needle on what matters most, it won't get the resources or attention to survive.
- Inadequate measurement frameworks mean organizations can't tell if their AI investments are actually paying off. Without clear metrics tied to workflow improvements, time savings or quality enhancements, AI projects remain experiments rather than becoming scalable solutions that warrant sustained investment.
Creating a framework for AI success
Understanding why AI projects often fail can inform how they are most likely to succeed. Designing frameworks that address each of the five most common failings, all while prioritizing strategic value over technological sophistication, will allow organizations to consistently achieve returns from AI investments.
Let’s explore some of these key framework elements.
Opportunity scoring provides the essential foundation for deciding which AI initiatives to pursue. Rather than chase every possible use case, evaluate opportunities based on potential impact versus implementation friction. High-impact projects with low implementation barriers become quick wins worth pursuing immediately. High-impact initiatives with significant implementation challenges may warrant strategic investment as part of broader modernization efforts. But low-impact initiatives requiring heavy implementation work should be avoided entirely – these drain resources and damage credibility without delivering meaningful returns.
Technology matching ensures the solution fits the problem at hand. Not every challenge requires generative AI or large language models. Some problems are better addressed through traditional machine learning, intelligent automation or even simpler rule-based systems. The key is selecting technology based on what the workflow calls for, rather than forcing the latest AI trend into every situation.
Workflow integration ultimately determines whether AI gets used or ignored in daily practice. This is why healthcare is moving towards more sophisticated orchestration approaches, such as AI agentification, where intelligent orchestrator agents coordinate specialized AI systems behind the scenes. This eliminates the need for users to interact with multiple disconnected tools.
These orchestrators interpret voice or text commands and delegate tasks across networks of specialized agents, each handling specific domains such as scheduling, documentation, revenue cycle management and patient engagement.
NextGen Healthcare's NextGen® Intelligent Agent, unveiled at the company's recent annual user group meeting, is an intelligent orchestrator agent that interprets voice or text commands and seamlessly coordinates a network of specialized AI agents – each with its own domain expertise. These agents autonomously execute tasks across clinical, financial and operational workflows, from retrieving patient insights and managing schedules to streamlining documentation and supporting revenue optimization.
The value of this type of framework is already proven. Ambient listening solutions reduce documentation and can save physicians around two and a half hours a day. Meanwhile, closed-loop patient engagement systems enable patients to schedule appointments through conversational interfaces – this also reduces administrative burdens and helps patients to be seen much sooner.
Building long-term, sustainable AI capability
Individual AI projects need to be bound together by several factors to create systematic organizational capability that can last. These include:
- Executive sponsorship: Senior leaders such as CTOs, chief data officers and/or chief AI officers should champion AI projects, securing the budget and organizational support needed to move initiatives forward.
- Governance committees: These teams set essential boundaries as AI capabilities expand. Healthcare AI governance must address fairness, transparency, patient privacy and regulatory requirements in addition to technical performance.
- Technical talent: The data scientists, AI engineers, and data engineers who build and deploy AI systems bring critical specialized expertise that keeps solutions running reliably at scale.
- Clinical informatics expertise: These professionals serve as translators between two worlds. They convert clinical workflows into technical specifications while helping engineers grasp the real-world complexities of patient care delivery.
- Project and change management resources: Strong project managers keep AI initiatives on track and on budget, and ensure stakeholders are apprised of progress, while change managers help clinical staff successfully adopt new AI-supported ways of working.
The path forward
AI's potential to transform healthcare is undeniable, and in many ways it already is. However, to realize the true potential of AI, organizations must move beyond enthusiasm and one-off projects to strategic implementation. Indeed, the organizations seeing returns from AI investments start with workflows rather than tools, prioritize integration over novelty, and measure impact rigorously.
The healthcare AI of the future will augment human expertise and make people’s jobs easier. Strategically implemented applications that work quietly in the background, freeing clinicians and staff to focus on delivering exceptional patient care, are key to meaningful success.
Learn more about strategic AI implementation in healthcare at www.nextgen.com.