Interview: Craig Allan Ahrens - Future of AI and workforce in Healthcare

Conversations on Artificial Intelligence, Healthcare, and Workforce

When I first met Craig, what stood out immediately was the breadth of his experience as part healthcare operator, part entrepreneur, part strategist, and part builder. His early AI-enabled clinical workforce platform helped hospitals act like their own staffing agencies, intelligently cascading shifts and giving central staffing teams an ability to flex, stabilize, and optimize the workforce

For more than 20 years, he has operated at the intersection of operations, workforce design, technology, and system strategy solving previously “unsolvable” operational problems inside major health systems in the U.S. and internationally:

As Craig said jokingly, “If it is boring, administrative, workforce, or operations related to healthcare - I am there.”

His experience advising health-tech founders, growing two healthcare startups through multiple investment rounds, teaching adjunct at graduate healthcare programs: innovation/operations/AI, creating some of the first AI healthcare operational startups, and working across both U.S. and EU healthcare systems has turned him into one of the rare translators between founders, investors, and healthcare innovators that are operating globally.

Today, Craig is once again at the leading edge in healthcare: building AI-native shared operational and workforce infrastructure that can scale globally, anchor workforce stability, and reshape not just who provides care, but how, when, and where it happens administratively. He calls it “AI Infrasharing”.

How do you see AI impacting jobs across healthcare systems globally?

In the next twenty-four months, AI should reshape 80% of roles and replace almost none. Reshaping is not the same as eliminating, and I believe that it will be a redistribution of labor, with, yes, some categories disappearing, but many new roles emerging.

For example, scribes become reviewers of AI drafts, nurses oversee AI-generated triage or education flows, and entry-level revenue cycle roles with repetitive workflows will see more direct displacement. However, the biggest impact is the elimination or substantial reduction of the administrative burden of healthcare.

That centers your work around AI Infrasharing. For readers unfamiliar with the term, what does it mean in the context of healthcare?

Yes, AI wrap-around intelligence, orchestration, and alignment layers, while the technology better attracts and engages workforces, stripping out administrative overhead. Infrasharing scales it and is the idea that health systems shouldn’t each build their own siloed AI, their own staffing workflow engines, or their own administrative automation stacks. Instead, they should share the infrastructure, not just the data:

1. Shared utilities rather than siloed point solutions;

2. Shared AI agents rather than shared data;

3. Shared administrative operational models that learn across systems without the sharing of sensitive competitive data;

AI Infrasharing turns workforce operations into a Workforce-as-a-Service AI model. Every system retains its autonomy and data governance, but they all benefit from the same AI-native rails:

· Vertical healthcare LLMs;

· Small, federated labor-data lakes;

· Shared orchestration engines;

· Standardized skill, credential, and job-architecture logic

The result is a massive reduction in administrative overhead in things like central scheduling, credentialing, compliance, recruiting, documentation, education, coordination, redeployment, etc. and a more flexible, resilient workforce across all of those who participate.

You’ve said AI will reshape 80% of roles without replacing most of them. How does Infrasharing accelerate or support that?

I am focused on outsourcing and automating the administrative load, using AI-driven orchestration to lift repetitive, low-value tasks off clinicians and staffing teams.

My work and projects are focused on pushing AI into the administrative dimension of the workforce or HR Tech with things like upskilling, scheduling, documentation, coordination of care, etc. so clinicians are freed up for the human part of healthcare delivery: listening, deciding, comforting, and leading teams. AI Infrasharing enables it and removes the biggest barriers to AI adoption and scaling across healthcare: fragmentation, inconsistent data, and every system trying to reinvent the wheel. When AI infrastructure is shared:

· AI features can be deployed faster;

· Training becomes standardized across the workforce;

· The administrative load is absorbed by a central intelligence layer.

New roles such as AI supervisors, coordinators, and reviewers can emerge consistently to foster the feedback loop with agents and evolve the orchestration exponentially.

For clinicians, it means fewer clicks and more care and for non-clinicians, it means AI handles the repetitive workflows so they can move into coordination, oversight, and problem-solving. AI becomes more equitable and better when it’s shared, not when every system attempts a bespoke build.

You’re now building an AI workforce infrasharing platform with a federated architecture. Why do you believe Europe is such fertile ground for this approach and how the US could learn from it?

Europe has three major advantages for enterprise-grade, federated AI in healthcare:

1. Strong public health systems with consistent workforce structures.

That makes standardized job architectures, shared skills taxonomies, and redeployment models far more feasible.

2. A cultural and regulatory foundation built on privacy and trust.

Federated learning thrives in environments where data must stay local. Europe is already comfortable with “data stays here, intelligence moves there.”

3. A willingness to collaborate at the system and national level.

Most EU hospitals already share learnings through networks. Federated AI is simply the next evolution. Europe can become the global reference point. This breaks the stereotype of how AI in Europe is stagnating. This is one area where it can teach the US how to scale rapidly.

It has the opportunity and there are early startups in the HR Tech space throughout Europe. Portugal specifically is becoming an ideal hub because the ecosystem is small enough to move fast but mature enough to be able to test pre-scaling across borders.

Let’s look at the human side. What is one task clinicians should stop doing, and one task AI should never control?

Clinicians should stop doing first-draft documentation. AI can draft notes faster, cleaner, and with fewer errors. Clinicians should be reviewers, not manual scribes.

AI should never control the final diagnosis and care plan for complex patients. Human judgment matters because medicine involves ambiguity, trade-offs, values, and context. AI informs, and humans decide. That’s the right balance.

A lot of people worry about a two-tier workforce emerging. Does AI Infrasharing change that calculation?

Yes, AI Infrasharing protects against a two-tier workforce. If every system builds its own AI, you get inequality: wealthy systems move fast, others fall behind. But a shared AI utility or AI Infrasharing means universal access to the same capabilities, shared training, shared upskilling paths and shared improvement cycles.

Upskilling and/or redeployment is the key foundation as AI improves and reshapes workflows, tasks, and roles. It will be the critical foundational layer for the next 3-5 years to truly unlock efficiency with AI to impact healthcare’s administrative burden.

What are the five skills that matter most in an AI-enabled health system?

Here are the top five:

1. AI & Data Literacy: Knowing what AI is doing, why, and where it fails.

2. Workflow Redesign: Teams must reshape the workflow, not build AI on top.

3. Agent and Workflow Orchestration: Understanding how to chain tasks across the people, workflows, and AI tools safely.

4. Risk Detection & Escalation: Knowing when not to trust the AI.

5. Learning Agility: Tracking how people adopt new tools and coaching them forward. Encouraging the innovation and adoption of “Shadow AI” out of the shadows.

If these skills aren’t in the job architecture, whether clinical and non-clinical, then the system can’t truly modernize.

You’ve worked at the intersection of startups, academia, and health systems for years. Why is that bridge-building role so important now?

Because AI Infrasharing forces all four key driving groups to work together:

1. Startups bring speed.

2. Health systems bring context, safety, and scale.

3. Investors bring discipline and repeatability.

4. Academia brings evidence and training.

Academia gets a bad wrap as being too conservative or a barrier to adoption. However, I disagree, and there are great groups like Stanford’s Human Centered Artificial Intelligence Institute that are leading the way in research and adoption. To me, AI Infrasharing is the meeting point, a shared place where ideas, evidence, and implementations converge. We can’t transform healthcare with siloed innovation anymore.

We need shared infrastructure, shared logic, and shared purpose in areas that healthcare systems can agree are not competitive advantageous zones; rather, baseline infrastructure that can be shared and enhanced with AI for shared overall efficiency and cost reduction. AI is just a tool, but AI Infrasharing is how we make the leap from isolated improvements to system-wide transformation in healthcare.

What makes Craig’s approach stand out is his rare ability to bridge two worlds. The urgency and creativity of the startup ecosystem, and the complexity and accountability of traditional healthcare. The goal? Not just to innovate, but to ensure measurable returns, improve patient outcomes, and drive a societal shift in how care is delivered and sustained.