AI technology and adoption are advancing at warp speed. Nearly every company in the Rallyday portfolio is adopting some form of technology to automate a clinical or operational function. We also have a front-row seat to the AI strategies of large enterprises, with two of our companies, Livefront and Nimble Gravity, supporting those transformation efforts directly.
One of our key observations is that organizations consistently underestimate the cultural and organizational work required to prepare for these tools. In healthcare, the challenge is even more acute because of complex workflows, unique revenue-cycle dynamics, and clinical implications.
For healthcare investors and operators, the implication is straightforward: AI should be evaluated not only as a software deployment, but as a change in how work gets done and how companies are run.
There are four reasons why preparation matters:
- Clinicians do not automatically trust new technology, especially after years of systems that added administrative work without giving time back to patients.
- Access is not enablement. Fewer than 25% of clinicians using AI tools today say they were adequately trained on them.
- The same technology produces very different outcomes depending on the workflow, governance, training, and feedback mechanisms around it.
- Investment is moving faster than measurement. Leaders need to prove where AI creates value before they scale it.
The leadership task is to address all four before expecting AI to deliver durable productivity gains.
Trust Is the First Adoption Barrier
AI is entering healthcare with a trust deficit.
Clinicians have lived through technology rollouts that were sold as improvements but often translated into more clicks, more documentation, and more time in front of a screen. The EHR was supposed to make care more connected and productive. For many clinicians, it instead became another source of administrative burden and another reason to spend time away from patients.
AI is not arriving on the healthcare scene with a blank slate. It arrives alongside the accumulated skepticism of every prior system that added work without delivering enough value in return.
Leaders who treat resistance as a communications problem will miss the point. Trust is not created by a launch email, a vendor presentation, or a promise that productivity will improve later. It is built through repeated experiences in which technology does what leaders said it would do.
Clinicians need to see that an AI tool reduces friction, fits the way care is delivered, and gives them more capacity for the human parts of the job. They need to understand how to review an AI-generated note, summary, recommendation, or message—and what happens when the output is wrong. They also need a credible answer to a basic question: what work will this tool remove or improve?
That is why trust must be treated as an operating metric, not a communications objective. If technology creates more exceptions, more review burden, or more uncertainty, adoption will stall no matter how compelling the business case looks on paper.
Training Is the Bridge Between Access and Value
The current training gap shows why tool activation is a poor proxy for adoption.
According to KLAS Research’s 2026 Arch Collaborative report, fewer than 25% of clinicians who have adopted AI tools agree that they received adequate training on how to use AI-generated content in their workflows. Among clinicians using ambient speech technology, those who strongly agree they know how to optimize the tool report an average Net EHR Experience Score of 89.7. Those who strongly disagree report a score of 46.7.
That is a 43-point gap between clinicians using the same technology.
A clinician can have access to an AI tool without understanding when to use it, how to review its output, what to do when it is wrong, or how the workflow should change. In that environment, AI does not feel like leverage. It feels like another responsibility added to an already overloaded day.
Training is therefore essential and must be practical, role-specific, and continuous. It should help clinicians practice the work they do, using real examples and clear review standards. It should explain where the tool fits, where it does not, and how feedback will improve the system over time.
A physician drafting an ambient visit note, a nurse summarizing a shift, and an operator managing patient messages may need entirely different forms of enablement. A physician-first rollout followed by a delayed, generic version for the rest of the care team will leave value on the table.
Perhaps most importantly, training cannot end when the tool is turned on. AI capabilities will change. Workflows will evolve. New risks will arise. Organizations need a repeatable process for measuring usage, gathering feedback, monitoring quality, and improving experience.
Workflow Discipline Determines Whether AI Multiplies Value or Friction
Trust and training are necessary, but neither can compensate for a broken workflow.
AI tends to amplify the environment it enters. In a fragmented organization, it can create more handoffs, more exceptions, and more confusion. In a disciplined organization with clear ownership and well-designed processes, it can multiply the impact of a strong team.
This is the central operational lesson: AI adoption should be treated as an operating-system change, not an IT project.
The organizations that capture durable value will do the unglamorous work around the tool:
- Map the workflow before automating it.
- Involve frontline clinicians in design and iteration.
- Standardize processes and clarify ownership.
- Define what good output looks like and who reviews it.
- Measure whether the technology reduces burden rather than simply moving work somewhere else.
- Continue coaching after deployment, when the real questions and failure modes emerge.
This is consistent with Rallyday’s broader view that healthcare has to fix its workflows before AI can fix them. Scaling a broken process is not a growth strategy. Adding AI to a process that clinicians already distrust is not a transformation strategy.
The goal is not to automate for the sake of automation. It is to make the work better for the people doing it and, ultimately, for the patients they serve.
AI Value Must Be Proven Before It Is Scaled
Adoption creates activity. It does not automatically create value.
Deloitte’s 2026 survey of 64 US healthcare CFOs and finance leaders found that organizations scaling AI were more confident in its financial potential, but less likely to report mature financial attribution capabilities. Only 18% of AI scalers said they consistently measured AI’s impact on revenue growth or cost savings with defined baselines and clear KPI ownership, compared with 31% of AI starters.
That is a meaningful proof gap. The organizations moving fastest may also be the organizations that find it hardest to separate AI’s impact from everything else changing in the business.
The challenge is not simply accounting. As AI spreads across workflows, multiple teams may shape the same outcome, baselines may shift, and benefits may show up indirectly through capacity, experience, quality, or avoided work. AI costs can also behave differently from traditional software costs. Usage-based consumption can rise with volume and complexity, which means a workflow may improve operationally while its financial return deteriorates if costs grow faster than benefits.
For operators, every significant AI use case should carry a simple value ledger:
- The problem being solved and the workflow being changed.
- The baseline performance before deployment.
- The operational and financial outcomes expected.
- The owner of each key performance indicator.
- The cost of usage and how that cost changes at scale.
- The cadence for reviewing results, risks, and whether to continue, adjust, or stop.
This discipline does not turn every benefit into an immediate P&L line. It does something more important: it creates a shared fact base for deciding what is working, what is not, and where additional capital should go.
Governance belongs in that same value equation. Human oversight, input and output controls, privacy, cybersecurity, and clear accountability are not brakes on progress. Done well, they create the confidence required to scale AI in workflows that affect patients, clinicians, and financial performance.
The Leadership Test for Investors and Operators
For healthcare investors and operators, the relevant diligence question is not simply: “Does this company use AI?” It is: “Has this company built the organizational muscle to use AI productively and prove the value?”
Before aggressively investing in AI and layering these tools into an organization, leaders should ask themselves:
- Do leaders understand the workflow they are trying to improve—and are these the right workflows?
- Have clinicians helped shape the implementation?
- Is training planned, measured, and reinforced?
- Can the organization clearly demonstrate that the tool reduces burden and improves performance?
- Is there a feedback loop for quality, safety, trust, and adoption?
- Does each material use case have a baseline, an owner, a cost profile, and a review cadence?
- Can the organization distinguish demonstrated value from expected value and activity?
The answers will increasingly separate companies that accumulate AI tools from companies that create real value with them.
After years of technology adding to the administrative burden, healthcare leaders need to prove that AI will be different by making work meaningfully better—not simply more digital. The leadership imperative is straightforward: prepare the organization before asking the organization to adopt the technology.
The healthcare organizations that lead in AI will not necessarily be the ones that move first. They will be the ones that carefully and thoughtfully prepare, measure, govern, and deploy.
Sources
- Deloitte, “Healthcare CFOs: Proving AI Value” — AI scaling, financial attribution, variable cost, and governance findings.
- KLAS Research, “AI Insights from the Arch Collaborative 2026” — training adequacy and clinician experience findings.
- Rallyday Partners, “Healthcare Has a Workflow Problem. AI Won’t Fix It Until You Do.” — Rallyday perspective on workflow redesign, frontline involvement, and durable AI value.
- American Medical Association, “Why Most AI Projects Fail in Health Care” — implementation and workflow considerations.
