This insight was featured in the April 16th, 2025 edition of the AI Catalyst Pulse.
Last week in Washington, D.C., nearly 100 CMIOs and CNIOs from leading health systems gathered to discuss AI implementation challenges, ambient listening technology, and nursing informatics priorities. Amid these discussions, we heard three distinct implementation strategies that stood out for their practicality and potential impact on healthcare's AI journey.
1. Earmark half of your short-term AI revenue gains for long-term transformation
It's always easier to justify investing in AI tools with immediate, measurable ROI – so it’s easy to find short-term “value-capture” AI applications, especially in the revenue cycle, dominating your AI strategy. The unintended consequence? While you chase short-term wins, you never build the foundation for genuine, sustainable transformation.
One technology leader proposed a simple “50/50 rule”: Whenever an AI revenue cycle tool generates financial gains, earmark half of those gains to fund longer-term transformational AI initiatives.
This creates a self-sustaining cycle of innovation: If you invest in an AI tool for CDI that gives you $4 million in new revenue, $2 million returns to the bottom line while $2 million funds more ambitious projects that might take years to deliver returns.
This approach requires pre-commitment. Once those revenue cycle tools start producing results, finance will naturally want to capture all the gains. By establishing the "half for transformation" principle before implementation begins, you can create a reliable funding stream for initiatives that might otherwise never get off the ground.
2. Pre-commit yourself – and your finance team – to success metrics before AI implementation begins
Stop me if this sounds familiar: Your clinical teams implement an AI solution and celebrate its success based on their metrics … only to have finance later declare it "wasn't worth the investment" using entirely different criteria.
One clinical informatics leader shared a simple approach to prevent this scenario: Have your leadership – and especially your finance leadership or CFO – explicitly define what success looks like before implementation begins.
This pre-registration of success metrics creates powerful alignment. Finance articulates specific ROI targets upfront, clinical leaders establish corresponding operational metrics, and everyone agrees on exactly what "success" means before spending begins.
The approach forces intellectual humility from technology enthusiasts by requiring them to demonstrate how their excitement will translate to measurable outcomes, while creating mutual accountability between clinical and financial stakeholders. It also requires people who are traditionally more removed from implementation challenges, such as finance leaders, to learn precisely what operational bottlenecks make it challenging to achieve KPI.
3. Think as if your AI contracts are 12-months long – even if they aren’t
Despite vendor pressure for multi-year commitments, leading health systems are doing their best to limit AI contracts to just one year. Why? Because in healthcare AI, both the technology and the competitive landscape are evolving at breakneck speed.
"Do not lock yourself into long multi-year agreements," advised one informatics leader. "The landscape is changing too rapidly."
This volatility is real. One health system told us they’ve replaced approximately one-third of its AI vendors in recent years. Many vendors in today's market didn't even exist three years ago, so it's naive to assume they'll all survive the next three.
And even if today’s market-leading vendor still exists in a few years, it may no longer be regarded as best-in-class. Consider ambient documentation: A few years ago, there were meaningful differences between vendors. Today, healthcare leaders feel the top technologies are all but interchangeable in standard outpatient settings.
So – where possible, negotiate short-term contracts. But what if you can’t – for instance, what if a vendor demands a three-year minimum commitment? We heard from several health systems who have found creative workarounds
Evaluating ROI on a 1-2 year basis even when forced into longer contracts
Including explicit language that you are not promising to implement products for the full lifespan of a contract
Evaluate upfront how you would switch away from a vendor’s products, including understanding your options to export critical data
Questions to consider:
Who in your organization should be involved in pre-defining success metrics for AI initiatives? How will you ensure these metrics balance financial, operational, and clinical perspectives?
If you earmarked a percentage of AI revenue gains for transformation, what would be the ideal allocation? Would a 50/50 split work in your environment, or would another ratio better reflect your strategic priorities?
What signals do you look for to assess an AI vendor's long-term viability? Beyond financial stability, how do you evaluate their product vision and ability to adapt to the rapidly changing healthcare landscape?