Before launching your next AI initiative, there’s a crucial question to ask: Is your data ready to support it? Without a strong data foundation, even the most advanced AI tools can amplify existing challenges rather than deliver meaningful value.
During a recent webinar hosted by Harmony Healthcare IT, Benjamin Cassity, Director of Research & Strategy at KLAS Research; Bill Russell, Founder and CEO of This Week Health and The 229 Project; and Jim Hammer, PMP, FACHDM, Chief Operating Officer of Harmony Healthcare IT, discussed what “AI-ready” data actually means and why it has become a prerequisite for successful AI adoption.
Here are some key takeaways from the discussion.
What AI-Ready Data Actually Means
While each panelist approached the question from a slightly different angle, all agreed on one fundamental point: AI is only as effective as the data behind it.
Russell, who previously served as CIO of St. Joseph Health, defined AI-ready data as information that is both accessible and well governed. “We have to know where it is, we have to put the controls around it, and it has to be accessible,” he said. “It also has to have structure and meaning once you find it.”
Hammer focused on data quality and consolidation. Organizations are often unaware of how many applications they have, where critical information resides, and how fragmented that information has become over time, he said. Understanding application inventory and improving data quality are essential first steps. “No one wants to feed an AI tool set or algorithm or LLM that’s incorrect,” Hammer noted.
Cassity built on those perspectives by emphasizing trust as the defining characteristic of AI-ready data. “AI-ready data is data that you can actually trust,” he said. “It needs to be complete, it needs to be clean, it needs to be accessible when you need it, and you need to know where it came from. A critical component of using AI safely is knowing the unknowns.”
Three Ways to Support More AI-Ready Data
Despite growing interest in AI, the panelists agreed that many healthcare organizations still face foundational data challenges.
“The good news is that the tools are getting better every day and the future of AI is really exciting,” Cassity said. “The hard news is that the data that is ultimately the foundational component of any AI strategy is usually very far from clean and very far from accessible.”
To close that gap, the panelists highlighted several areas of focus for healthcare organizations seeking to strengthen their data foundations.
1. Establish Clear Data Governance
AI-ready data requires more than technical accessibility. Organizations need clear standards for how data is defined, managed, validated, and maintained over time, said Russell.
Just as important, someone must be accountable for ensuring those standards are upheld. “Someone has to be able to ask the right questions, validate what they’re seeing, and have a mechanism in place to clean those things up and improve over time,” he said.
2. Reduce Data Fragmentation
Many healthcare organizations still have data spread across EHRs, departmental applications, and other systems throughout the enterprise. Before organizations can fully leverage AI, they need a clear understanding of where their data resides and how it is connected, said Hammer. If data is fragmented and siloed, initiatives to consolidate and enhance data access and quality may be necessary.
“A lot of times we run into teams that aren’t aware of all the applications they have, or of how siloed data is across the organization,” he said. “Rationalizing that inventory, understanding what’s duplicative, enhancing the cleanliness of that data, is critical.”
3. Enhance Data Integrity and Trust
Even accessible, consolidated, and well-governed data can create problems if organizations do not understand its origins, completeness, and limitations. Cassity emphasized that healthcare leaders must be able to trust the data being used to support AI initiatives.
“AI-ready data is data that you can actually trust,” he said. “It’s got to be complete. It’s got to be clean. It’s got to be accessible when you need it. You have to know where it came from.”
Final Takeaway
While the principle of “bad data in, bad data out” is not new to healthcare, the scale and speed at which AI can amplify data quality issues is changing the game. Organizations that understand their data lineage, validate data quality, and address inconsistencies before deploying AI will be much better positioned for successful initiatives.
“If you fix nothing else, get your data clean and make sure it’s trusted and manageable,” Cassity said. “You get that right and AI becomes a genuine accelerator.”
This is the first article in a series drawn from a recent Harmony Healthcare IT webinar on AI-data readiness. Next up: The Role of Legacy Data in AI-Readiness.Sidebar:
Industry Perspective: What Hospital IT Leaders Are Saying about AI-Ready Data
According to the 229 Project’s “Signal Index Spring 2026 Report,” based on a survey of 146 hospital IT leaders, “data credibility, access, and governance remain the precondition for AI and operating intelligence.” The report also found that healthcare organizations increasingly recognize that data strategy and AI strategy must be addressed together.
The three webinar panelists echoed this sentiment throughout the discussion, emphasizing that trusted, accessible, and well-governed data is foundational to successful AI initiatives.
