MTI Viewpoints

Insights shared by industry relative to healthcare and the advancement of medical technology.

Jaimon Jose, senior director of engineering and compliance at CharmHealth, is a software architect and technical leader with more than 18 years of product development experience in the areas of identity and security technologies, distributed systems, cloud security, and data center virtualization.

Technology has promised to transform healthcare for years, but so far, the results have been underwhelming. Instead of ease and efficiency, we got more clicks, screens and administrative burdens.

This moment feels different, however.

According to Menlo Ventures’ 2025 State of AI in Healthcare report, healthcare providers now spend over $1 billion annually on AI adoption. This figure alone is striking, given healthcare’s traditionally measured response to new technology. Yet, we’re suddenly seeing health systems move from cautious pilots to production deployments with a sense of urgency, and our industry is setting the pace for enterprise AI adoption. The shift behind it is notable: Agentic AI, autonomous systems capable of reasoning, planning and taking action, is here.

Agentic AI fundamentally changes what is possible, even in an industry as highly regulated as healthcare.

The Cause for Excitement

Agentic AI introduces capabilities that go well beyond prior generations of tools. Instead of prompting users to do something, agentic AI actually does it — and not just in a siloed application but across entire workflows. This distinction matters because healthcare workflows are rarely linear. A single patient interaction can involve intake, documentation, diagnostics, billing and follow-up, each with its own systems and requirements. When agentic AI is put to work within this maze, it makes functions seamless across the entire patient journey. This creates a more cohesive, less error-prone experience. In short, impact becomes meaningful.

Agentic AI connects these previously disparate pieces, in part, because it understands context. Based on its contextual understanding of each function, agentic AI can determine whether it should act, moving users through the workflow, or pause to defer to human judgment. In doing so, agentic AI accomplishes what few previous technologies have been able to accomplish: It eliminates friction rather than adding to it.

Importantly, the agentic AI opportunity does not lie with a single vendor or solution. It lies in the infrastructure, the orchestration of all of the individual pieces in concert with one another. Open, interoperable standards can enable this type of innovation to flourish across the entire healthcare ecosystem. That’s where emerging open frameworks such as the model context protocol (MCP) and similar interoperability standards come into play.

MCP and Emerging Standards: The Foundation for Healthcare AI Agents

One of the biggest historical barriers to innovation thus far has been fragmentation. Data is siloed across electronic health records (EHRs), billing systems, scheduling platforms, and external databases, and it has been really difficult to bring these systems together in a simple way.

Frameworks like MCP are helping change this. They provide a method for AI systems to connect external tools and data sources. McKinsey’s healthcare AI analysis notes, “Open architectures, such as MCP, are allowing new AI agents to directly access functional data across care organizations.” Essentially, MCP is what gives AI agents superpowers. In practice, approaches like MCP can help simplify the highly complex integration challenge of healthcare workflows. Instead of building one-off connections, developers use a common framework to enable AI systems to do things like access patient records, check drug interactions, schedule appointments, or process claims.

And developers have responded. In just one year, MCP has gone from an open-source experiment to an industrywide standard. MCP grew from roughly 100,000 server downloads when it launched in November 2024 to over 8 million by mid-2025, with more than 5,800 MCP servers now available. Then last December, MCP was donated to the newly formed Agentic AI Foundation under the Linux Foundation, co-founded by Anthropic, OpenAI, and Block, ensuring vendor-neutral governance and long-term stability.

With the foundation in place, we now can completely reimagine the patient journey, starting with the very first patient touchpoint through post-visit actions like insurance claims and follow-up.

With the introduction of standardized approaches for running interactive user interfaces through MCP, AI systems have been increasingly able to present structured data, dashboards and decision workflows within the context of care delivery. Think visualization of patient trends, guided forms for documentation, or approval pathways for clinical decision, each dynamically generated and context-aware.

Through these advances, AI agents are becoming a routine part of daily practice, not just another layer of applications that clinicians and administrators are forced to navigate.

The Human Element Remains Central

As exciting as these developments are, the role of human judgment remains vital. No amount of data analysis or algorithmic precision can replace human expertise, accountability and empathy.

To clarify, the purpose of agentic AI is not to replace healthcare providers but to support them by alleviating their cognitive and administrative burdens, allowing them to concentrate more on patient care. This fosters a shared responsibility among technology providers, healthcare organizations, and patients.

This framework ensures that developers and technology providers make systems that are safe, transparent and auditable, while healthcare organizations carefully integrate these tools into their workflows. Ultimately, providers must maintain the authority to make final decisions on diagnoses, treatments and patient care plans; every prescription, diagnosis and treatment requires human approval. Patients also have a role to play. The success of AI-supported care depends on their willingness to share accurate and complete health information, including family history, current medications, allergies and symptoms. When patients actively participate with their care team, the AI systems operating behind the scenes — such as capturing interviews and analyzing data — can deliver more personalized and accurate support.

Humans, at every stage, are at the heart of this model.

The New Look of Healthcare

The convergence of agentic AI, open interoperability standards, and healthcare’s pressing need for efficiency creates a significant opportunity. In fact, industry analysts predict that agentic AI in healthcare will grow 40-45% annually, with the potential to exceed $5 billion within the next five years.

If this trajectory continues, the tension between technology and care delivery may finally begin to ease as agentic AI fulfills its potential to improve life for clinicians and patients.

The question for healthcare systems, EHR vendors, and care delivery organizations is no longer whether agentic AI can reshape clinical workflows. It’s whether these stakeholders have the faith to embrace the promise of another technology for a shot at real transformation.