Healthcare AI has moved quickly from concept to tool set, and Epic customers especially feel that shift. Recently, KLAS interviewed leaders at 30 unique organizations about the Epic AI tools they are using and where they have seen the most and least impact. We also asked for their views on pricing and how well Epic is supporting adoption. The feedback is broadly encouraging with some caveats.
Overall, customers like that Epic moves quickly and gives them options. Many are seeing practical wins, particularly in workflows where the task is concrete and the outcome can be measured. At the same time, the interviews make it clear that simply turning on AI is not the same as winning with AI. We see that success comes when organizations invest in workflow alignment, user education, validation, and measurement.
Points to Know
- Epic’s AI portfolio creates real impact across clinical and administrative workflows, but value varies by use case and organization.
- The same tools are being cited as being impactful and not impactful, showing that adoption and fit matter as much as functionality.
- Customers generally describe Epic’s AI pace and support as positive, but many want stronger change management and adoption guidance.
- Pricing is acceptable for most customers today, but confidence will depend on clearer ROI, utilization visibility, and predictable costs as usage scales.
Customers Are Seeing Value, but Not Uniformly
When asked which Epic AI tools have been most impactful, respondents most often cited In Basket Drafts and Outpatient Insights. Denials Appeals, Coding Assistant, Inpatient Insights, AI Charting, and a mix of smaller tools also received positive feedback. This breadth suggests that additional workflow-specific capabilities could continue to reduce manual work in different parts of the organization.
The strongest praise tends to come when the workflow is concrete and the value can be counted. One director described measurable output across multiple areas:
“We’ve had over 880 generated summaries for 11,000 notes, saved 18,000 hours with some revenue cycle automation, and drafted 3,400 appeals through AI, which has resulted in a 17% reduction in the time we would typically spend creating those.” —Director, February 2026
While most respondents did not have such specific metrics on hand, all agreed that AI conversations are shifting from “AI is interesting” to “Where is AI producing measurable operational or clinical relief?”
When AI Disappoints, It Is Often a Workflow and Adoption Problem
The interview data also shows why early AI wins can be fragile. In Basket Drafts and Outpatient Insights are both frequently named as impactful tools and as least-impactful tools. When looking at why tools weren’t seen as impactful, reasons include individual organizational or departmental differences such as the initial configuration, specialty-specific workflows, and how much effort customers put into helping users understand why and when to use the tool.
In Basket Drafts is the clearest example. Some organizations see value because drafts reduce manual message work. Others report that clinicians spend too much time editing responses, especially when patient questions are complex. Outpatient Insights has a similar pattern. Summarization can be useful, but specialists may find the output too generic when they need focused, specialty-specific context.
This points to a practical lesson for Epic customers: Not all AI tools are plug and play in the way a simple feature may be. Most tools require validation, tuning, specialty-aware workflow design, and direct communication about what users should expect. Leaders who report having poor first experiences with a tool are generally still optimistic about its future potential, but when end users lose initial trust, it can create reluctance for future adoption even when the product improves later.
Customers Want Epic to Keep Moving, but They Need Help Keeping Up
Most respondents describe Epic’s pace of AI development as about right, albeit a stretch to keep up with. Customers know AI is moving quickly across healthcare, and many would rather have new options to evaluate than wait for a slower road map. One CMIO said, “It’s fast. I’d say it’s about right though, because the industry is moving at a certain pace. I can say it’s too fast, but I’d rather have options than be waiting on them.”
Still, fast development alone doesn’t guarantee quick results. Customers say they need better visibility into which AI tools are being used, clearer guidance on what to validate and monitor, and more support helping clinicians absorb AI-driven workflow changes.
Epic’s support is generally viewed positively, with most respondents saying the vendor has provided the support they need. But the comments behind the scores show that support is not always consistent across tools, with some reporting bottlenecks with certain Epic staff. For some AI capabilities, customers feel they are building their own playbooks for governance, training, change management, and value measurement.
Pricing Is Acceptable Today, but the Future is Less Certain
For now, most interviewed organizations say Epic has an appropriate AI pricing model, which is important given the breadth of offerings Epic has and how sensitive pricing has become across the market. However, the positive sentiment comes with a clear qualifier: Many customers are still early in adoption and in lower-cost periods, and they are trying to understand how suite pricing will feel after utilization grows.
“The first year is much cheaper than year two, so we have to see the value. Today, yes. Ask me next year. If I don’t see the value, I might say differently.” —Director, February 2026
To decide whether the pricing remains fair once AI is no longer a pilot or in an introductory bundle, customers want better visibility on the usage and benefits of AI tools. They also want better forecasts on the budget implications of scaled use, and where possible, the financial return they should expect.
AI Pricing Is a Marketwide Challenge, Not Just an Epic Conversation
Broader KLAS commentary about paying for AI solutions from other vendors shows that provider organizations are starting to push back on AI monetization when the cost feels unpredictable, difficult to scale, or disconnected from the value being delivered.
Unsurprisingly, most comments about AI pricing are negative with customers citing high overall price and added fees required to use the tool. More positive comments appear when customers see clear ROI, when pricing is predictable, or when licensing makes it easier to scale adoption. Per-user pricing is often a sticking point because it can discourage broad use, while enterprise or flat-fee approaches are viewed more favorably because they give organizations budget certainty.
Epic’s suite pricing helps to lower the barrier to trying many AI tools, but over time, customers expect alignment between what they pay, the tools they use, and the value they receive.
Looking Ahead
Epic has a real advantage in customer AI adoption. Many customers want to have an Epic-first approach because the EHR is already the center of their workflows and the integrated tools can reduce the burden of managing yet another vendor. But an Epic-first approach is not the same as an Epic-automatically approach. Customers keep third-party options alive when those tools are more specialized, faster moving, or better supported with ROI and adoption materials.
Looking ahead, customers will judge Epic’s AI progress less by the number of tools released and more by whether customers can repeatedly move from implementation to adoption to outcomes. The customers are willing to move quickly; they just need clearer evidence that the speed produces lasting value.
About the Data
KLAS interviewed leaders at 30 unique Epic customer organizations about the Epic AI tools they are using, the tools they view as most and least impactful, the support they receive, the pace of Epic’s AI rollout, and AI pricing. The AI pricing context is based on KLAS commentary analysis from provider organizations using non-Epic AI-enabled solutions. Sample sizes vary by question because not every respondent answered every question.
© LALAKA / Adobe Stock
