CAIO’s First 90 Days in Healthcare: What Works, What Fails

I have had several conversations about AI leadership over the last few months. I removed the specific examples below, and I recognize the role of AI and Technology Leaders are continuing to evolve. Many companies and organizations are hyper-focused on jumping on the “AI train”, but there is an opportunity to approach this more successfully and avoid snake-oil.

Introduction

The Chief AI Officer (CAIO) role in healthcare is rapidly emerging and evolving, demanding a leader who is not only a visionary but also deeply pragmatic, capable of navigating complex ecosystems and delivering real value quickly under high pressure. Lives are on the line, and regulators are watching closely. While the allure of groundbreaking AI innovations is strong, a CAIO’s first 90 days are critical. This is when foundations are laid for long-term success or failure. A successful healthcare CAIO should move through a three-phased plan—Learn, Assess, Execute—to build credibility, de-risk deployments, and set the foundation for sustainable AI transformation across payer, provider, and health-tech. The takeaway: a CAIO must be more than a visionary; they must be a pragmatic leader who delivers quick wins responsibly, grounded in governance and enterprise strategy from day one.

Structured For Success

Success hinges on establishing robust governance, validating existing structures, and demonstrating scalable outcomes. Echoing lessons from the CIO role. A structured 30-60-90 day entry plan—focused on People, Process, and Technology—accelerates impact: roughly 30 days learning, 30 days planning, 30 days executing early wins. This deliberate pacing guards against “do everything at once” and builds credibility through understanding and planning before execution.

Days 1–30: Understanding the Business & Building Trust

Get the lay of the land: Meet stakeholders across the enterprise such as administrators and clinicians (provider), claims directors and medical directors (payer), and product/engineering (health tech). Inventory existing AI initiatives, both successful and stalled, including “shadow AI,” and map AI and data across the enterprise. Critically, validate whether Responsible Use of AI governance operates in practice: active review and approvals, production guardrails and monitoring (bias, drift, privacy, security).

Engage and reassure the workforce: Healthcare organizations mix excitement and anxiety about AI. Front-line staff might worry about job displacement or algorithmic decisions impacting care or their workflows. The CAIO’s role is as much change management and trust as technology. Don’t arrive as the hotshot tech guru dictating solutions before listening. Emphasize how AI eases burdens (e.g., documentation) or identifies at-risk patients, rather than implying replacement. Evaluate the current AI/data org, capabilities, and platforms (e.g., cloud, data platforms, LLM/ML providers).

Days 31–60: Strategy, Governance, and Early Wins

Formulate an AI strategy: Select one or two lighthouse use cases with lower risk and high ROI. Examples like reduce clinical variation (providers) or improve growth/experience (payers). Explicitly decide what not to pursue (too risky, misaligned with regulation/values). Showing the C-suite that the CAIO is focused on business value, not just shiny tech.

Governance: Governance & guardrails will make or break AI in healthcare. Borrowing a racing metaphor shared by Charlene Li: “if you want to go fast, you need really good brakes.” Establish or evaluate a Generative AI Center of Excellence and a Responsible AI framework; standardize intake, evaluation, approvals, and production monitoring. Choose evaluation/red-teaming/monitoring tooling. There should be no CAIO silo, this must be a cross-functional team who surfaces roadblocks and celebrates wins. Move governance from paper to practice with clear sign-offs and controls (e.g., EHR-embedded models documented for intended use, performance, bias mitigation, and risk management across validity, reliability, fairness, robustness). By day 60, AI should be innovative and safe, not the Wild West.

Days 61–90: Execute, Deliver, and Communicate

Deliver the win: Launch secure pilots/POCs with tracked outcomes: turnaround time reduced, coding errors caught, cost savings, or time returned to caregivers. Quantify impact and openly share lessons. Avoid declaring victory too early; credibility comes from transparent, practical delivery.

Upskill and empower: Stand up an AI upskilling program with literacy tiers (general staff, power users, technical). In healthcare, train clinicians on clinical decision support or contact-center teams on AI assistants (when to rely on them and when to override). This signals to the organization that the CAIO isn’t just installing fancy software, but that they are investing in the workforce to succeed in an AI-enabled future.

Communicate relentlessly: A new CAIO must also become the organization’s AI storyteller and evangelist (in an authentic, non-hype way). Present to the C-suite and broader staff; share regulatory updates and internal progress to maintain momentum and position the CAIO as the org’s trusted AI voice.

Conclusion: Pragmatic Visionary

The first 90 days set the tone for whether AI becomes a transformative asset or a distrusted experiment. Borrowing from the CIO’s evolution, the CAIO must blend vision with pragmatic execution and guardrails: think big, start small, move fast—safely. What works is aligning AI to purpose, people, and prudent strategy; what fails is chasing AI for its own sake or without a safety net. Just as CIOs mastered the art of managing infrastructure and applications, CAIOs must master the intricacies of AI development, deployment, and ethical governance within a complex healthcare environment. The CAIO of today can learn from the CIOs of yesterday: be the bridge-builder and change agent who turns innovation into practical value. If you’re standing up the CAIO function, I’m happy to compare notes.