“We Need an AI Strategy” Is the Most Expensive Sentence in Healthcare Right Now

An executive says, “We need an AI strategy,” at the board meeting or on a call with a funder. It sounds responsible. It sounds forward-thinking. Every peer organization seems to be saying some version of it. And it almost always triggers the same next steps: a request for proposals, a consulting engagement, and a standalone AI strategy document that costs somewhere between $75-$250K.

We see this pattern across healthcare organizations of every size in California. The sentence itself is not the problem. AI is reshaping healthcare workflows and organizations that do not prepare will fall behind. The problem is the response the sentence triggers. When AI gets its own strategy, it gets its own lane. And that lane almost always runs parallel to the strategic plan the organization already operates.

What Happens After the Strategy Engagement

A consulting firm conducts stakeholder interviews, benchmarks what peer organizations are doing, and delivers a polished, 60-page document. It includes a landscape analysis, a readiness assessment, and a list of recommended AI use cases. It references the organization’s strategic plan, but it doesn’t modify it. It proposes new governance structures instead of expanding the ones already in place. It recommends AI tools without mapping them to the specific workflow steps those tools would actually live.

The document goes into a shared drive folder. Leadership discusses it at a couple meetings. And then the organization faces the question the document was supposed to answer but never quite did: what do we actually do with it?

Across industries, 70-85% of AI projects fail, not because the technology was wrong, but because of implementation gaps. In healthcare specifically, a 2025 Black Book Research survey found that only 8% of organizations that adopted AI documentation tools reached positive ROI within the first year. Only 10% of healthcare organizations had formal AI oversight in place. The gap between strategy and operations is where most AI investments stall.

The Separate Lane Problem

The most expensive consequence of a standalone AI strategy is not the consulting fee. It’s the parallel governance structure it creates.

When AI strategy lives in its own document, AI governance sits outside existing compliance. A new AI committee forms, separate from the compliance committee that already meets monthly. AI vendor purchases are reactive to sales pitches because the strategy document does not connect to the procurement review the organization already runs. Staff figure out AI on their own because the strategy is written for the C-suite, not for the care coordinator managing a documentation backlog or the billing specialist reviewing denial codes.

Two parallel systems now compete for the same leadership attention, the same committee time, and the same staff bandwidth. The strategic plan says one thing. The AI strategy says something adjacent. And the people doing the work follow neither, because neither document tells them what to do at 2PM on a Wednesday when they notice a new AI feature in the EHR.

This is not a criticism of the organizations that commission these documents. The buying cycle moves faster than the oversight cycle. The pressure to act on AI is real, and a standalone strategy feels like the answer.

The Question That Costs Less & Produces More

The alternative is not to ignore AI. It’s to change the question. Instead of “What is our AI strategy?” ask “Where does AI strengthen the strategy we already have?”

That question leads to a different set of actions.

For strategy, it means opening the existing strategic plan and identifying the 3-5 goals where AI makes existing organizational commitments stronger. If the strategic plan prioritizes reducing documentation burden, the AI question becomes: which documentation workflows take the most staff time, and which steps within those workflows could an AI tool improve? The answer lands inside the strategic plan itself, not in a parallel document.

For governance, it means expanding the scope of the compliance committee the organization already runs. Add AI as a standing agenda item. Track AI tool performance through the same operational dashboards the committee already reviews. Investigate AI incidents through the same incident reporting process teams already use. No new committee. No new reporting line. The governance structure stays consolidated, and AI oversight gets the same rigor the organization already applies to HIPAA, Joint Commission, or HRSA compliance.

For procurement, it means expanding the existing vendor review. The organization already evaluates vendors for data handling, security, and regulatory compliance. Adding AI-specific questions to that existing review covers the gaps: Does the tool retain data after the task ends? Does the vendor use organizational data to train its models? Can the organization opt out? Does the tool produce different results for different demographic groups? Has the vendor disclosed known limitations? Same review process, broader scope.

For teams, it means anchoring training to specific workflows, not to AI as a concept. The care coordinator does not need a lecture on large language models. She needs to know that the AI-generated visit summary requires her review before it enters the medical record, that she has override authority when her professional judgment says the output is wrong, and that she can report errors through the incident process she already uses. An additional 30-minute module added to the annual HIPAA training she already completes covers this.

You Already Have the Foundation

If you operate a strategic plan with measurable goals, you already have the structure to fit in AI planning. If you run a compliance committee that meets regularly, you already have the governance body that can oversee AI. If you conduct vendor reviews, you already have the procurement process to evaluates AI tools. If you train staff annually on HIPAA and data privacy, you already have the training pathway that covers AI-specific guidance.

AI does not need its own lane. It fits inside the lanes you already built. The organizations that will use AI most effectively are not the ones with the thickest strategy documents. They are the ones that ask the right questions, map the right workflows, and expand existing structures they already trust.

The most productive version of “We need an AI strategy” turns out to be: “We need to figure out where AI fits in the strategy we already have.”

Has your organization been through a standalone AI strategy engagement? We’d love to hear what happened after the document was delivered.

 #HealthcareAIStrategy #HealthcareAIGovernanceFramework #AIGovernance  #AIReadiness  #TrustworthyAI #HealthcareAIROI