Partner Spotlight: Q&A with Asher Informatics
19 November 2026
Insights from John Kalafut MSEE PhD, Founder and CTO at Asher Informatics
Tell us about Asher Informatics and your role there.
Asher Informatics helps healthcare delivery organizations govern AI across its lifecycle, from deciding whether to build, buy or partner through deployment, monitoring and retirement. We cover predictive, generative and agentic AI, because non-generative AI systems carry real risk too.
The company grew out of Asher Orion Group, the consulting firm my co-founder Charlotte and I launched in 2021, where many clients struggled with the same question: how do you actually deploy AI inside a health system?
In late 2023, as ambient AI and chatbots spread across healthcare, we shifted from a services-led model to a platform-first approach. As co-founder and Chief Technology Officer, I lead product and solution strategy, drawing on 30 years in medical technology and health IT.
How did Asher Informatics become involved with CHAI?
We've followed CHAI since 2022, when it was a group of clinicians and data scientists recognizing the need for a consortium to set shared expectations for AI. When CHAI released its blueprint in 2023, we began engaging more directly, and I've served on CHAI working groups since 2024. Becoming a partner was a natural next step, and we’re excited to work together in this new capacity.
AI adoption is accelerating across healthcare. What challenges is that creating for health systems?
Health systems face growing pressure to adopt AI. Boards and executive teams see its potential to improve efficiency and stay competitive, but service-line leaders are often left asking what they actually need, how to determine what value is achievable, and how their ways of working will or need to change with the new AI-enabled technology. As with the first wave of radiology and cardiology AI around 2018, the core challenge is separating signal from noise and not just determining an AI model’s statistical performance. Healthcare organizations need to know if an AI system or application will be useful and generate value to the users, organization, and patients.
At the same time, requests keep multiplying, from agentic platforms to AI embedded in the EHR. Each deployment expands an institution's risk surface, and more mature organizations, particularly in imaging and those with tools embedded in EHR systems, are asking how to monitor what they've deployed and measure its value.
Agentic AI adds another layer. Clinical care will likely keep a human in the loop, but agents are already entering operational areas of the hospital, bringing new requirements for monitoring and oversight. As we have seen with self-driving technology, just saying “a human can override” can give a false sense of security. The users interaction with their task is changed, and that human computer interaction is a particularly nuanced risk surface in high-risk environments - like healthcare.
Why is AI governance becoming such an important priority for healthcare organizations? What are those that are successful doing to approach governance?
AI systems aren't conventional software. A model operates within a broader system of interfaces and workflows, but its behavior isn't always repeatable, which makes it harder to integrate and oversee.
Accountability is also a major driver. For regulated AI products, product liability offers some clarity and precedent. But if an internally built agent makes an error that disrupts accounts receivable, responsibility is far less clear. Organizations need to understand who is using these tools, how they can fail, how failures will be detected and how to demonstrate oversight to auditors, regulators, communities and patients.
The most effective approach we see is federated: a core governance group tied to executive leadership sets direction, while domain specialists evaluate the use cases, consider failure modes because they are closest to their work. These organizations also connect governance to existing procurement and risk assessment processes. That works for a 200-bed community hospital as well as an academic system, and CHAI's governance playbooks help put it into practice.
How has collaborating through CHAI shaped your perspective?
I've served on four CHAI working groups, and one discussion in the clinical decision support group illustrates the value especially well. A physician shared that a maternal-infant risk algorithm embedded in their EHR wasn't producing useful results. Another participant explained that the tool had been designed only for at-risk infants, something the clinical team had never been told.
That exchange shows why multidisciplinary perspectives matter. A device engineer may refer to "indications for use," a physician may associate the phrase with a drug label, and a software developer may not recognize it at all. CHAI creates a trusted environment where those perspectives come together and participants feel comfortable sharing what isn't working, not just what is.
What advice would you give to others considering partnering with CHAI?
Bring the people closest to the problem. Commercial teams have a role, but the greatest value comes when product managers, engineers and other subject matter experts participate directly. Working groups are a place to solve real challenges together, not to promote a particular solution.
It's also worth recognizing that CHAI welcomes industry participation. People who build health technology are generally motivated to improve care, and the coalition is stronger when developers and providers work side by side.
Looking ahead to 2027, why does this moment matter?
For years, many organizations planned to address AI governance once it became necessary. With growing litigation, evolving federal regulation and new state laws, that point has arrived. Some hospitals now aim to review, select and deploy 100 to 250 algorithmic tools a year, a scale challenge that can't be managed without strong governance.
CHAI also helps counter playbook fatigue. What I mean by that is there are many resources developed by many groups and organizations available. Everything from free excel sheets to SKILLS for AI agent systems developed by commercial and community groups. As organizations develop their own approaches, CHAI offers a common foundation that brings community and academic health systems, developers and other stakeholders together. It’s also important to stress that the methods used by CHAI Working Groups to assemble the best practice playbooks and other artifacts is akin to the same rigor and approaches for generating clinical guidelines and engineering standards. Because health systems look to their peers, that shared, national perspective carries real weight.
At Asher Informatics, our platform, FORGE™, applies the risk-based, evidence-based rigor of regulated engineering to AI governance without adding unnecessary burden for clinicians. Having been part of CHAI since its early days, we're proud to continue contributing to that work as a partner.
