< + > Human in the Loop Does Not Mean Safe: The Hidden Risks of Agentic AI in Healthcare

The healthcare industry treats artificial intelligence like just another software rollout. We write policies, form committees, and assume standard guardrails will keep operations safe. But agentic AI behaves entirely differently than previous digital health tools.

At the recent eHealth26 conference in Halifax, Nova Scotia, I sat down with Julia Zarb. She is the force behind Blue x Blue, a company building AI infrastructure with governance incorporated. She warned that applying the old IT playbook to this new medium creates massive unseen risks.

Core Insight: Human Oversight is a Liability Trap Slapping a “human-in-the-loop” label onto an AI tool does not guarantee patient safety, it simply shifts the legal and operational liability onto overwhelmed clinicians.

The Human-in-the-Loop Fallacy

Executives love hearing that a human will review AI outputs before finalizing a clinical decision. In reality, these reviewers lack the time and system explainability to catch microscopic errors under high stress.

“What we’re doing is transferring the liability that comes with that review,” Zarb noted. “We don’t have what’s called explainability to go back in and see how that got to it, who knew what, when, and where, and why?”

The Threat of Invisible Errors

AI models fail in tiny, logical steps that easily bypass human reviewers.

“What makes agentic AI good is also its internal flaw,” Zarb explained. “The way that it processes, it carries forward little micro fissures and flaws.”

Standard IT guardrails fail because AI naturally jumps boundaries. When multiple AI agents interact, a single logical error propagates until it results in a much larger failure.

“If 100 agents speak to 100 other agents and those mistakes are brought forward, these aren’t mistakes that are just errors, these are mistakes that make sense [to Ai at the time] because of the way the AI is built.”

Questions Healthcare IT Leaders are Asking

Why is the “human-in-the-loop” approach risky for healthcare AI? Relying solely on human oversight transfers immense liability to clinical staff. Fast-paced environments and high stress prevent humans from properly auditing AI outputs. Furthermore, clinicians lack the technical explainability tools required to understand exactly how the AI arrived at its conclusions.

How do AI errors differ from standard digital health software bugs? Traditional software bugs are often clear and trackable. AI generates “micro fissures” or small gap-fillers that look perfectly logical to the system. These tiny errors bypass standard guardrails and multiply rapidly when different AI agents share data across an organization.

Learn more about Blue x Blue at https://www.bluexblue.com/

Listen and subscribe to the Healthcare IT Today Interviews Podcast to hear all the latest insights from experts in healthcare IT.

And for an exclusive look at our top stories, subscribe to our newsletter and YouTube.

Tell us what you think. Contact us here or on Twitter at @hcitoday. And if you’re interested in advertising with us, check out our various advertising packages and request our Media Kit.

< + > Applying AI to Healthcare Integration and Interoperability

One of the coolest AI tools I’ve seen recently was the healthcare integration agent, Rhapsody Axon.  It’s an AI integration agent embedded directly in the Rhapsody and Corepoint platforms, built to speed up the work integration teams do every day. It also lines up with my theme that AI is going to be able to be applied to almost every aspect of healthcare.  In fact, that’s what makes Axon so cool.  Rhapsody applied AI to a topic that they understand deeply, and they know is a challenge for the interoperability and integration experts out there.

Here’s what I keep coming back to: as more AI agents show up across healthcare, they’ll all eventually need to move data across systems. Axon is an early example of an agent that doesn’t just answer questions about integration, it can actually help build the connections (not just wait for a human to do it).

This is why I was really excited to check out their webinar called “What’s Next in AI for Healthcare Interoperability? See Rhapsody Axon in Action.”  Click the link to watch the full webinar or check out the highlights below that we shared on social media during the event.

What do you think about applying AI to healthcare integration and interoperability?  What other novel approaches to AI in healthcare have you seen?  Let us know on social media.

How does Rhapsody Axon reduce the grind of integration work?

I’ve heard this from a number of integration experts in the past.  Many of them spend more time looking at the documentation than they do actually coding.  Not to mention time spent troubleshooting an integration that has broken because some interface has changed.  The time spent on these actions are real and required.  Axon’s AI helps remove some of this friction.

Is Rhapsody Axon embedded or a standalone tool?

I think every CIO I talk to is afraid of AI sprawl.  In other words, they’re afraid of having hundreds of different AI applications that they have to support.  Thus, it’s a welcome sight that Axon is embedded in Corepoint and Rhapsody.

How does Axon understand your integration environment?

I’m also impressed that Axon not only understands the standards and integration documentation, but it also understands your Corepoint and Rhapsody set up.  This type of personalization to your environment is so powerful and helpful when building new integrations or troubleshooting existing ones.

How does Axon help onboard new integration staff?

Of course, I also love the idea of Axon quickly upskilling a new person at your organization rather than having the new person continually bothering your other integration staff.  A new employee can ask Axon all the questions they want without annoying them.

How does Rhapsody Axon handle data security and privacy?

These are the types of questions I see a lot of CIOs asking of their AI vendors when it comes to AI security and privacy.  I love that Rhapsody just shared these details up front.

I think we’re probably going to have this conversation about technology replacing humans forever.  Clearly humans have shown themselves to be pretty resilient.  It’s great to see Axon helping those humans be more efficient.  As I often say, AI is going to replace all those time consuming things that humans don’t want to do.  That seems to be what Axon is doing for integration experts.

Can Rhapsody Axon build integrations automatically?

I was a little surprised that searching through standards documentation wasn’t on the list of things that most integration experts spend time on.  Although, maybe that’s part of the two that they listed.

No doubt these two tasks are important and accelerated by AI.  As their customer said on the webinar, if you don’t like Axon after you see how it can automatically create the integration routes for you, then he doesn’t know what to tell you. These agents can consume interface specs and create integrations for review. I’d add that if you don’t see the value in agentically created integration routes, you’re probably not an integration engineer.

AI programming engines have become all the rage.  Programmers have really come around to using them.  The problem is they don’t work for this kind of custom coding that’s needed for healthcare integrations.  That really is what makes Axon unique, but still in line with what’s happening with coding in other languages.

I imagine that an integration engineer’s least favorite task is upgrading their integration from one HL7 standard to the new one, or migrating an old HL7v2 interface over to FHIR. That’s a bunch of work to essentially make sure it keeps doing the same thing.  That’s not fun work.  In other words, it’s perfect work for an AI agent.

How are teams using Axon to review integration documents?

This example from Jennifer of how they’re using Axon was really brilliant.  No one likes reviewing documents for a new integration.  Axon can do that quickly and highlight the areas that really matter so you can quickly have the right conversation with the vendors you’re looking to integrate with.  That’s powerful.

I’ve felt Jennifer’s pain a lot with the Healthcare IT Today website.  I know just enough to be dangerous and things that would be easy and quick for an expert take me far too long.  AI has really helped me solve these more technical issues quickly.  Sounds like she’s doing the same in her job with Axon.

Having Axon run reports for her is fascinating too.  I love the concept that she ran the reports manually for so long that she can really know if the AI is doing well or not.  The fact that she trusts the AI now is saying a lot.

Those are a few of the highlights from the webinar.  You can watch the full webinar including a demo of Axon or check out the Rhapsody Axon page to see it in action for yourself. What do you think of this use case for AI in healthcare?

< + > This Week’s Health IT Jobs – July 29, 2026

It can be very overwhelming scrolling through job board after job board in search of a position that fits your wants and needs. Let us take that stress away by finding a mix of great health IT jobs for you! We hope you enjoy this look at some of the health IT jobs we saw healthcare organizations trying to fill this week.

Here’s a quick look at some of the health IT jobs we found:

• Business Development Representative – The Garage
• Senior Data & AI Engineer – The Garage
• Coding Client Success Specialist – WellSky
• Program Consultant, Health Care – American Heart Association
• Appeals and Grievances Specialist – Medica
• Business Systems Analyst – Dept of Medicine – Hartford HealthCare
• Patient Service Specialist – Texas Children’s Hospital
• Medical Billing Office Representatives – Concentra
• Patient Access Representative – Patient Access Center – CHRISTUS Health
• Medical Records Reviewer – VITAS Healthcare
• Chief Medical Coding and Compliance Officer – XpertDox
• Director IT Portfolio & Program Delivery – Regulatory & Strategic Programs – CVS Health
• Associate Chief Digital Information Officer – AMN Healthcare Leadership Solutions | B.E. Smith
• Sr. Director – Enterprise Connectivity & Network Services – Eli Lilly and Company
• Market IT Principal Consultant and Team Lead – Kaiser Permanente
• Head of Data & Infrastructure – Crimson Health
• Manager II Technology – Elevance Health
• HIM QA Lead, Remote, Health Information Management – Baptist Health
• Medical Device Cybersecurity Risk Specialist – Tata Consultancy Services
• Deputy CIO, Care Technologies & Informatics – Information Solutions – MUSC Health

If none of these jobs fit your needs, be sure to check out our previous health IT job listings.

Do you have an open health IT position that you are looking to fill? Contact us here with a link to the open position and we’ll be happy to feature it in next week’s article at no charge!

*Note: These jobs are listed by Healthcare IT Today as a free service to the community. Healthcare IT Today does not endorse or vouch for the company or the job posting. We encourage anyone applying to these jobs to do their own due diligence.

< + > Lessons from CHIMA: Coding for Clinical Care Instead of Reimbursement

The Health Information Day at eHealth26 in Halifax was an unexpected treat. The event hosted by CHIMA felt entirely different than previous revenue cycle and health information management (HIM) conferences. The biggest surprise was the lack of discussion on payer friction and claim denials.

Core Insight from CHIMA26

The conversations in Halifax proved that health information management is moving out of the back office. Without the constant battle of fighting insurance companies for reimbursement, Canadian professionals are focusing increasingly on data quality and governance. They are actively evolving into the essential data stewards needed to make future AI projects successful.

Coding for Care Instead of Cash

US hospitals code every minor patient condition to justify costs and ensure payment. In Canada, the single-payer model changes the rules entirely. Jodi McMullin, founder of ScoJo Consulting, holds certifications in both countries and explained the stark contrast.

“Because it’s a funding portfolio, we don’t look at those smaller issues,” McMullin explained. Because primary care in Canada is a capitated system, physicians are not as concerned about reimbursement accuracy. As long as the patient is treated and the information recorded into the EHR, everyone just moves on.

Expanding the Profession

The health information role is expanding rapidly across Canada. Mahmoud Suliman, CEO of CHIMA, pointed out that the profession often struggles to explain its true value to the public. He wants to change that narrative.

“There are a good number [of our members] that are coders, but we’re finding there are people part of clinical documentation improvement, release of information,” Suleiman explained.

These professionals ensure patient data remains secure, accurate, and accessible for clinicians and policymakers making critical decisions for the health system.

The Bottom Line from CHIMA’s Health Information Day

Health information management is no longer just a back-office administrative task. These professionals are strategic assets for data governance. Organizations looking to establish a strong and accurate data foundation, need look no further than their HIM department for a pool of capable professionals ready for the challenge.

Questions Healthcare IT Leaders are Asking

Why is clinical coding volume lower in Canada than the US? The United States operates on a multi-payer system that forces providers to document every condition to secure reimbursement. Canada uses a single-payer system that focuses strictly on the primary reason for the patient visit. This results in fewer codes and far less administrative friction.

How do health information professionals impact AI strategy? Artificial intelligence models require accurate, clean data to produce safe clinical outputs. Health information professionals act as data stewards who enforce governance and maintain privacy compliance. Their oversight guarantees that the data feeding these models is trustworthy.

Learn more about CHIMA at https://www.echima.ca/

Learn more about ScoJo Consulting at https://scojocoding.com/

Listen and subscribe to the Healthcare IT Today Interviews Podcast to hear all the latest insights from experts in healthcare IT.

And for an exclusive look at our top stories, subscribe to our newsletter and YouTube.

Tell us what you think. Contact us here or on Twitter at @hcitoday. And if you’re interested in advertising with us, check out our various advertising packages and request our Media Kit.

< + > The Future of Value-Based Care Will Require Actuarially Engineered AI

The following is a guest article by Brian Overstreet, President & CEO at Arbital Health

At a recent Summit on the Future of Value-Based Care (VBC) and Risk Contracting, more than 150 payer executives, provider leaders, policy influencers, and digital health innovators gathered to discuss the operational realities of managing risk. While the attendees represented organizations deeply committed to value-based care, their conversations revealed concern that the industry’s ambition for aligning financial returns to better patient outcomes is advancing faster than the infrastructure that supports that goal.

Three key themes emerged from the discussions at the event:

1. Leaders are confident that VBC is working but are still having difficulty showing direct financial impact
2. The first step in making value-based care succeed is actuarially scoring the value of clinical interventions so that the economics are clear
3. AI’s value, cost, and long-term impact on VBC is still relatively unknown

Showing Direct Financial Impact

Despite momentum in the evolution of VBC, many organizations still struggle to measure results, particularly within a one-year contract cycle. Delayed performance data, misaligned measures, and slow operational processes make it hard to show financial impact on demand. Forecasting and risk adjustment are still not moving fast enough to support the industry’s ambitions.

Too many organizations are still relying on information cycles built for fee-for-service: lagged claims, quarterly refreshes, and retrospective reporting. In most risk contracts between payers and providers, that approach is no longer good enough.

The organizations gaining ground are not waiting for a settlement to understand performance. They are using actuarial AI to track performance trends against contract dynamics in real time, surface cost drivers earlier, and identify interventions before negative trends become financial results.

The First Step: Actuarially Scoring Clinical Interventions

Participants at the Summit described a practical formula for success in specialty value-based models:

1. Actuarially score the value of clinical interventions, so the economics are clear
2. Create meaningful provider incentives and engagement
3. Deliver high-quality clinical care that improves patient outcomes
4. Design benefits that reduce friction and drive patient participation

The first step—scoring interventions using actuarial logic—proved particularly important as it sets the stage for financial success, without which the rest becomes unsustainable. Too often, programs are launched with clinical enthusiasm but limited financial clarity and early measurability. Only actuaries can help set value against the counterfactual effects of VBC, i.e., the positive outcomes that didn’t happen (deaths avoided, etc.), versus those that did happen. Without actuarial modeling of potential outcomes, even well-intentioned and successfully executed interventions can struggle to demonstrate value within short contract windows.

The Debate Around AI

Healthcare organizations are investing heavily in AI, yet many executives are now asking if AI actually reduces costs, or if it is simply adding another layer of expense. Infrastructure investments, integration challenges, staffing requirements, and model maintenance can quickly accumulate for organizations trying to build AI solutions internally. And, for organizations already managing thin margins, the financial return on AI investments remains under scrutiny.

That skepticism is shaping how leaders evaluate new technologies. Tools that promise predictive insight are increasingly expected to deliver both operational efficiency and measurable financial impact.

A Shift Toward Actuarial AI

Most healthcare AI tools are built by data scientists trained to optimize predictions. What’s different about Actuarial AI is that it is built by actuarial engineers and trained on actuarial-grade logic — credibility theory, loss ratio modeling, risk adjustment mechanics, trend analysis, reserving principles, and regulatory guardrails – all tied to individual contract terms and measures. It does not “black box” outputs, but surfaces its assumptions, model logic, data sources, and calculations to be fully transparent and auditable.

That distinction matters because in risk contracting, trust is currency. If finance leaders cannot audit the model, they will not rely on it. At the Summit, participants discussed the shift from standard reporting to accountable analytics, that is, systems that show their math.

The Strategic Implications

VBC will continue expanding across Medicare Advantage, ACO REACH/LEAD Model, Medicaid managed care, and commercial risk arrangements, and the attendant complexity will increase. Organizations that pair actuarial expertise with actuarially engineered AI will price better, see performance trends faster, and provide actionable insight into bending the risk curve so that organizations can manage risk more confidently.

Industry leaders are optimistic about what comes next for VBC, both from a policy standpoint and in how it operates day-to-day, but they recognize that success will depend on more than expanding risk contracts.

Organizations seeking to move ahead will use AI to drive immediate efficiency and faster decision-making, while requiring transparency and auditability. The next phase of VBC will be defined not just by risk contracts, but by the intelligence infrastructure that supports them.

About Brian Overstreet

Brian Overstreet is the Co-Founder, President, and CEO at Arbital Health. Brian has over twenty years of experience working with data and analytics SaaS companies in the healthcare market.

< + > Corner Health Announces $32.5 Million in Seed and Series A Financing | TytoCare Names Adam Pellegrini as CEO and Closes $25M+ Growth Round

Check out today’s featured companies who have recently raised a round of funding, and be sure to check out the full list of past healthcare IT fundings.

---

Corner Health Announces $32.5 Million in Seed and Series A Financing to Help Independent Clinicians Rebuild Primary Care

Corner Health, a company designed to rebuild the healthcare system by empowering Nurse Practitioners (NPs) to start and scale their own local primary care practices, announced today it has raised $32.5 million in Seed and Series A financing. Their recent Series A was led by Oak HC/FT, with participation from existing investors First Round Capital and Zigg Capital. The investment follows a period of rapid growth for Corner Health, which has established a network of more than 70 provider-owned primary care practices. Corner Health is the fastest-growing primary care network in Arizona and Washington and has enabled 35,000+ patient visits in the past year.

The funding comes at a pivotal moment for healthcare. Despite unprecedented investment in AI, more than 100 million Americans still lack access to a primary care provider, clinicians are burning out at record rates, and patient visits continue to get shorter. Corner Health is taking the opposite approach: using AI to create more time for the provider-patient relationship, not less.

“I grew up watching my mom run her own private practice, and I saw firsthand how powerful it is when a clinician owns the relationship with their patient,” said Lava Sunder, Co-Founder and CEO at Corner Health. “The tragedy is that the economics of healthcare have made that model increasingly difficult to sustain. Healthcare has spent decades pushing clinicians into larger and larger systems, often at the expense of time with patients. Via Corner Health’s network, the most common word that appears in patient reviews is ‘listen.’ When providers have more time, patients feel heard.”

The backbone of Corner Health’s model is Cora, an AI-native practice operating system designed to automate the full spectrum of brick-and-mortar primary care operations, including scheduling, patient communication, billing, lab orders, referrals, and prior authorizations. Built for scale and in-person clinical environments, Cora enables 90% of Corner Health clinics to operate with no additional staff, dramatically reducing overhead and giving NPs more time to focus on patient care.

“The dominant use of AI in healthcare today is helping clinicians move faster and see more patients,” said Anne Gifford, Co-Founder and COO of Corner Health…

Full release here, originally announced July 13th, 2026.

---

TytoCare Names Adam Pellegrini as CEO and Closes $25M+ Growth Round to Scale AI-First Clinical Enablement Platform

Company Repositions as an AI-First Clinical Enablement Platform, Bringing FDA-Cleared AI-Powered Insights into Virtual Primary Care to Support Cardiopulmonary and Oncology Care

TytoCare, pioneer of remote physical examination technology featuring FDA-cleared medical devices, today announced two major developments: the appointment of Adam Pellegrini as Chief Executive Officer, and the closing of a $25 million-plus growth round led by Insight Partners along with HOOP, OliveTree, OrbiMed, Qumra Capital, Qualcomm Ventures, and others. The company is expanding its platform to deliver clinical-grade remote care for patients with chronic and complex disease, positioning TytoCare to redefine how virtual primary care is delivered across high-acuity chronic disease populations.

Pellegrini brings more than two decades of experience at the intersection of digital health, consumer health technology, and large-scale clinical programs. His appointment comes as TytoCare moves to embed its remote examination platform and FDA-cleared AI-powered SaMD (software as a medical device) algorithms directly into integrated care pathways for remote cardiopulmonary monitoring, oncology support, and complex chronic disease management, disease areas where the gap between in-person clinical rigor and virtual care delivery has remained wide.

TytoCare’s platform combines a handheld examination platform featuring FDA-cleared medical devices — capable of capturing clinical-grade heart, lung, ear, skin, throat, and abdomen data — with a suite of FDA-cleared AI-powered SaMD algorithms that enable clinicians to conduct comprehensive remote physical exams with diagnostic confidence previously achievable only in person. The company’s expanded clinical enablement strategy is set to deepen integrations with leading health systems, payers, and employer health programs, with particular focus on congestive heart failure (CHF), COPD, post-surgical recovery, and oncology treatment monitoring.

“The convergence of a clinically validated exam device, AI-powered diagnostic algorithms, and the urgent demand from health systems for real clinical intelligence at the point of virtual care is an extraordinary and rare combination,” said Adam Pellegrini, Chief Executive Officer. “TytoCare has built the foundational infrastructure for the next generation of intelligent remote care, and I could not be more energized to lead this next chapter.”

The oversubscribed financing round, led by Insight Partners with participation from existing strategic investors, reflects institutional conviction in TytoCare’s market position and the accelerating commercial demand for AI-enabled remote diagnostics…

Full release here, originally announced July 15th, 2026.

< + > The Hidden Risk of Rushed Technology: Patrick Lo on Establishing AI Governance

The healthcare industry is rushing to adopt artificial intelligence tools, but many organizations are moving too fast without proper safeguards. Deploying automated systems without a clear plan for risk management can expose a health system to major compliance liabilities and operational failures.

Healthcare IT Today sat down with Patrick Lo, CEO of Privacy Horizon, at the eHealth26 conference to discuss why healthcare organizations must establish solid AI governance before deploying new digital health solutions.

Core Insight: Trust Before Intelligence

Healthcare organizations must establish comprehensive AI governance, including cross-functional risk management frameworks and clear lines of accountability, before deploying any automated clinical or administrative tools.

The Necessity of Pre-Deployment Controls

Healthcare leaders rush into vendor contracts without evaluating the long-term operational impacts on their staff. Patrick warns that organizations must prepare their compliance infrastructure well before any software integration goes live. “You just cannot just deploy a solution and hope for the best without thinking about what should prepare and put in place before you adopt a solution,” Lo explained.

Establishing Committee Accountability

Effective governance means bringing different stakeholders to the table to validate the operational purpose and risk profile of a new tool. Committees must assign clear ownership to handle system errors and manage vulnerabilities before technology enters the clinical environment.

“Before you deploy anything that’s AI, you should have an AI governance, which means including how do you make decision to deploy an AI? Why do you need to deploy an AI? Who should be accountable?” Lo noted.

Questions Healthcare IT Leaders are Asking

Why is AI governance required before a technology deployment? AI governance establishes critical rules for decision-making and risk management prior to software integration. Without this foundation, healthcare facilities risk implementing tools that compromise patient privacy or lack clear lines of staff accountability when automated errors occur.

What primary questions should an AI governance committee address? A cross-functional committee must explicitly define the clinical or administrative problem the technology is intended to solve. Leaders must also determine who is operationally accountable for the data inputs, system outputs, and ongoing risk mitigation frameworks.

Learn more about Privacy Horizon at https://www.privacyhorizon.com/

Listen and subscribe to the Healthcare IT Today Interviews Podcast to hear all the latest insights from experts in healthcare IT.

And for an exclusive look at our top stories, subscribe to our newsletter and YouTube.

Tell us what you think. Contact us here or on Twitter at @hcitoday. And if you’re interested in advertising with us, check out our various advertising packages and request our Media Kit.