AI at AI Speed | Healthcare AI Decisions Beyond AI Model Selection

Lee Akay

Over many years of working alongside healthcare organizations evaluating and implementing emerging technologies, Innovation Discovery Center (IDC) has observed that enduring changes in organizational management rarely become visible through a single project. They begin to emerge when similar operational questions appear repeatedly across different organizational settings.

Individual healthcare organizations naturally experience technological change one initiative at a time. Broader organizational patterns become visible only after those same operational questions begin recurring across many organizations and workflows.

As AI becomes embedded throughout healthcare operations, we are beginning to observe exactly that kind of pattern.  Clinical teams are evaluating AI to assist with documentation. Revenue cycle leaders are exploring coding and reimbursement support. Patient access teams are assessing tools to improve scheduling and communication. IT departments are reviewing AI capabilities embedded within the enterprise software they already use.  While these initiatives address different operational needs, many organizations are encountering remarkably similar management questions regardless of the workflow being considered.

How should AI solutions be evaluated?

How should competing approaches be compared?

How should success be measured?

How should today's decisions inform tomorrow's investments?

What initially appears to be a series of independent AI projects is gradually becoming something else: a recurring organizational challenge that extends across the enterprise.

Every week seems to bring another announcement about new AI capabilities. Frontier models continue to improve. AI is becoming embedded within enterprise software. Specialized solutions continue to emerge for clinical documentation, revenue cycle, patient communication, imaging, scheduling, and other operational workflows.  Against that backdrop, one question naturally emerges: Which AI solution should we choose?

Selecting an AI solution remains an important decision. As organizations evaluate AI across more operational areas, however, they often discover that choosing a solution is only the beginning. The larger challenge is establishing a consistent way to make AI decisions over time.  Organizations soon find themselves asking many of the same questions regardless of the vendor or application being evaluated.  How should AI be evaluated within our organization?  What operational criteria matter most?  What governance should be established before deployment?  How should performance be measured after implementation? How should experience gained from one AI initiative improve the next?

These questions begin to shift the discussion from selecting individual AI solutions to developing a repeatable approach for making AI decisions.

Healthcare organizations have successfully navigated similar transitions before. Medical devices are evaluated through structured clinical, operational, financial, and regulatory processes.  Enterprise software undergoes governance before implementation. Imaging technologies are assessed based on clinical need, operational fit, integration requirements, financial considerations, long-term support, and measurable value.

Healthcare has developed disciplined approaches for evaluating technologies that influence patient care and organizational performance. Artificial intelligence appears to be following a similar pattern.  There are, however, a few important differences.  Healthcare organizations have always adapted to changing technologies, but AI raises new questions around cost and data sovereignty, including how consumption-based pricing compares to traditional licensing, and what happens to proprietary data once it is shared with a model provider that may one day compete in adjacent markets.  The most persistent difference, however, is the cadence.  Capabilities continue to evolve at a pace measured in months rather than years, requiring organizations to revisit decisions far more frequently than they have with most traditional healthcare technologies.

The impact of that accelerated cadence is unlikely to be uniform across the enterprise.

Administrative functions may experience it first not because they are more important, but because the barriers to change are different.  AI supporting documentation, revenue cycle, scheduling, patient communication, and other operational workflows can often be evaluated and deployed more rapidly than technologies directly affecting clinical decision-making.

Clinical applications are likely to follow a different rhythm. Patient safety, clinical validation, physician acceptance, regulatory expectations, and governance appropriately create a more deliberate adoption process. Although the pace of adoption may differ across the enterprise, the underlying management challenge remains remarkably consistent. Organizations are being asked to make AI decisions more frequently, across more operational areas, and with greater confidence than ever before.  That shift has important implications for AI pilots.

A well-designed pilot should certainly evaluate technical performance. Technical performance alone, however, rarely determines operational success. The more enduring value of a well-designed pilot is that it helps an organization develop a repeatable approach for evaluating AI within its own operational environment.

How does the solution perform inside existing workflows?

What governance questions emerge?

What operational changes are required?

How easily does it integrate with existing systems?

How should success be measured?

What should be monitored after deployment?

What lessons should guide future AI initiatives?

These questions often outlast the evaluation of any individual AI solution because they become part of how the organization approaches AI more broadly.

Healthcare organizations have repeatedly demonstrated an extraordinary ability to adapt as medicine, technology, reimbursement, and regulation have evolved.  Electronic health records introduced new governance and operational processes.  Value-based care required new approaches to performance management and organizational coordination. Population health expanded how organizations measured outcomes and managed care across larger patient populations. Each technological or operational shift ultimately required corresponding changes in how organizations managed their work.  Artificial intelligence appears to represent the next stage of that progression.

As AI becomes part of everyday operations, organizations are beginning to develop new approaches for evaluating, deploying, governing, and continuously improving these capabilities across the enterprise. The organizations that benefit most from AI may not ultimately be those that simply identify the best models or software. They may be the organizations that deliberately develop the management capabilities required to evaluate, deploy, govern, and continuously improve AI as technology continues to evolve. Those capabilities will likely outlast any individual model, platform, or vendor because they become part of how the organization learns, adapts, and improves over time.  Based on what IDC is observing through operational AI deployment across healthcare organizations, that transition is already underway. AI at AI speed requires management systems redesigned to match the cadence of the technology itself.