AI in Healthcare: What’s Ready and What’s Not

Healthcare conferences in 2026 share one thing in common: nearly every session mentions AI. Brett Cooper and Lee DeHihns from BlueFletch attended several healthcare events over the past four months and estimate that 95% of presentations included an AI segment. The hype is real, but the question hospital IT teams need to answer is more specific. Which clinical and operational workflows are genuinely ready for AI today, which ones are not, and what risks emerge when clinicians start adopting AI tools on their own?

Watch the full episode for the complete discussion.

What Clinical Tasks Is AI Ready to Handle Right Now?

AI is ready for administrative, repetitive, and time-consuming tasks, not for those that require clinical judgment, diagnosis, or human empathy. The clearest use cases fall into three categories: documentation, billing, and transcription.

On documentation: physicians routinely report spending 3 to 4 hours of a 10-hour shift charting in EHR systems instead of engaging with patients. AI-powered ambient listening and dictation tools let clinicians record notes, chart medications, and log procedures while maintaining eye contact with the patient. DeHihns notes that two different physicians recently pulled out personal devices during appointments to use AI transcription for real-time charting. The consensus among healthcare professionals is building quickly around AI for documentation.

On billing: every procedure, medication, and interaction generates codes for insurance reimbursement. AI can automate billing code entry, flag missing codes, and cross-reference claims against payer requirements. For hospital systems where billing and insurance management overhead is substantial (including internal staff, insurance company processes, and third-party auditing firms), automating revenue cycle workflows with AI offers one of the fastest paths to measurable efficiency gains.

On staffing: predictive models that analyze historical patient volume data can help hospital operations teams forecast staffing needs six to twelve months out. For organizations already stretched thin, data-driven staffing decisions reduce overtime costs and improve coverage during peak census periods.

What AI is not ready for: clinical judgment, diagnosis, and the human empathy that patients need from their care teams. “I don’t see it as a replacement necessarily just yet,” DeHihns says.

Prefer to listen? Catch this episode on your favorite podcast app.

What Is Shadow AI and Why Should Hospital IT Teams Worry?

Shadow AI is the healthcare version of shadow IT, and it’s happening now. Clinicians are bringing personal AI tools into clinical settings to solve problems their employer’s systems don’t address. A physician using an unapproved transcription app on a personal phone during a patient encounter is a real scenario, not a hypothetical one.

The risks are significant, especially with shared clinical devices, where multiple clinicians access patient records during a single shift. An unapproved AI tool may not be HIPAA compliant. Patient data entered into a personal AI instance could be stored outside the hospital’s control, used to train public language models, and leaked through a security gap the hospital has no visibility into. “Where is that data being stored?” DeHihns asks. “What’s to stop a nurse or a clinician from copy-pasting patient information into their own personal instance of whatever AI engine they prefer?“

A blanket ban on AI is not the answer. Clinicians are problem solvers by nature. If a tool is not available from their employer, they will find their own. The better approach is a governance framework that acknowledges AI usage, designates approved tools with enterprise privacy settings, and communicates clear policies. “Here’s what’s approved. Here’s how we have enterprise settings set up so that our data is kept private,” DeHihns says. “Let people know that yes, this is out there. Yes, you should be using it. But here’s a way to do it safely.”

BlueFletch’s own nursing technology survey found that better training on existing systems would ease clinician burden. Adding AI without adequate training support risks creating the same frustrations that plagued early EHR rollouts.

How Should Hospital IT Leaders Prioritize AI Investments?

Hospital budgets are tight. According to Kaufman Hall’s National Hospital Flash Report, roughly half of U.S. hospitals operate at a loss or break even, with rural and safety-net hospitals facing the steepest shortfalls. For CIOs evaluating where to invest in AI, the advice from this discussion comes down to three principles.

1. Start with the problem, not the technology.

“You don’t buy a technology and look for problems to solve with that technology,” DeHihns says. “It would be like buying a screwdriver and walking around your house for everything that the screwdriver fit.” Identify a specific operational pain point (documentation time, billing errors, staffing gaps), test an AI tool against that problem. Measure the before-and-after results.

2. Run small pilots before enterprise rollouts.

No healthcare CIO has spent their entire career as an AI expert because the technology is too new. Start with a focused proof of concept, measuring results against revenue cycle improvements or staffing efficiencies. Validating outcomes before expanding reduces the risk of premature deployment. If you try to do everything at once, the rollout will fail the same way poorly planned EHR deployments did.

3. Listen to your frontline clinicians.

Doctors and nurses are the end users. Their feedback on which tasks consume the most non-clinical time, which tools they find cumbersome, and where they are already seeking workarounds tells you exactly where AI investment will deliver the fastest return. Rolling out AI without clinician input leads to the same adoption problems hospitals have seen with every prior technology wave.

Frequently Asked Questions

Your clinicians are already using AI, whether your hospital has approved it or not. Book a demo to see how BlueFletch helps healthcare organizations secure shared clinical devices while keeping workflows fast for nurses and physicians.