Eliminate the Hype: 3 Things to Know Before Adopting Agentic AI

The healthcare industry has entered the “age of agentic AI,” marked by a rapid increase in the number of new AI startups and significant funding. Yet 80% of health systems report lacking resources to identify, select and implement AI solutions.

So here’s the problem: health systems that could benefit the most from agentic AI are often very slow or even unable to adopt it.

For over a decade, I have seen several “epochs” of health IT evolution. “Agentic AI” is the next solution to help solve what healthcare has been focused on for years: connecting patients to care easily and efficiently, while reducing staff workload.

Yet what is new and different is the profound, and currently untapped, potential of AI to completely transform the way patients interact with healthcare, including the administrative and communications side of patient access – scheduling, rescheduling, referrals, prescriptions, post-discharge, general questions and more.

Here are three considerations for healthcare organizations to make when adopting an agentic AI solution for patient communications:

Don’t be fooled by “lightweight integration” capabilities

Every vendor will promise “deep” EHR integration because that’s the foundation of a truly valuable AI agent. The ability to securely read and write to patient records is how an agent automates workflows and provides real value.

But how do you know if a vendor is a true leader in this space, and not just “getting by”? A key indicator is their adoption of new, forward-looking standards.

For example, consider the Model Context Protocol (MCP). It is a new open standard developed by Anthropic that fundamentally changes the way AI agents interact with external systems. Think of it as an “API for AI agents” that fills the gaps in traditional APIs compared to LLMs, enabling more secure, dynamic, and efficient interactions with systems like EHRs.

Are your potential partners actively adopting MCP? If so, do they build a simple wrapper or do they think about the task end-to-end and design the MCP function to limit the LLM’s judgments? Evaluate their team and their investments: do they have the technical foresight and resources to do more than just get by? The right partner will not only meet today’s integration needs, but also enable your healthcare system to take full advantage of future advancements.

Avoid the miracle solution trap

When adopting a new solution, leaders want a long-term, sustainable solution, not a temporary one. Yet many new implementations fail to deliver lasting value because they only solve the surface problem, leaving the fundamental issues, at the infrastructure level and manual work, unresolved.

I’ve recently heard about agentic AI implementations that still rely on file transfers or manual data entry. Although this approach results in rapid deployment, it is a quick fix in a worst-case scenario. This perpetuates manual processes, increases the risk of using outdated data and/or the risk of mishandling PHI/PII, and fails to free up administrative staff in the long term.

In contrast, a true agentic, autonomous AI solution uses dynamic integrations into downstream systems (e.g. via MCP) or fits naturally into a broad portfolio of solutions already offered by the vendor. These approaches provide a sustainable return on investment and may involve a slightly longer initial implementation.

Additionally, resist the temptation to over-engineer agent prompts for a quick fix, as some vendors can fit everything into an agent prompt in the service of rapid implementation. Although this solution may seem effective for rapid commissioning, it is fragile and presents risks. While carefully designed intent-based MCP tools can increase performance, reduce the risk of hallucination, and improve scalability.

There are also questions that need to be answered to help provider organizations anticipate the future: will this implementation still be useful in six months or a year? Does today’s focus on speed sacrifice the deeper, more transformative value your system deserves tomorrow? These are important considerations as healthcare providers weigh the value of speed in terms of sustainability and security.

Avoid the security “check box” mentality

There is no “finish line” when it comes to security and privacy, especially in the age of agentic AI; the landscape is constantly changing. When I started in this industry, I thought HITRUST certification was enough. Today I know that is not the case. Such certifications are only a snapshot and not a reflection of an ongoing commitment to protecting your healthcare system’s most sensitive data.

While many vendors are taking the right steps – obtaining certifications, hiring security managers, and implementing standard protocols – this should be the starting point. Health data security and privacy have evolved far beyond what they were just a few years ago, and now agentic AI systems – which constantly learn, adapt and make decisions on their own – are making this situation worse. As agentic AI introduces new security challenges, healthcare leaders should prioritize partners that not only have strong security measures in place today, but are also actively committed to staying ahead of AI security trends and can quickly adapt to new threats with effective solutions, such as mitigating data leaks using MCP or implementing testing agents to analyze and score conversations.

Today, a strong commitment to security and privacy must be fundamental and cultural. A security-first mindset and commitment must extend across technology, departments, and people across the enterprise. It is an organizational value, not a handful of certifications managed by a small team responsible for “governing” the rest of the employees.

For example, companies working with government agencies can apply for FedRAMP High authorization, the U.S. government’s most rigorous security standard. It’s a smart move because it’s more than just checking a box. The process itself further embeds a culture of safety throughout the organization.

In conclusion, in the new era of agentic AI, you don’t just buy a piece of technology; you’re adopting a system that will learn, adapt, and become part of your team’s operations. An agentic AI provider should be a partner as invested in your long-term success as you are.

How to tell the difference? A vendor sells you a solution and gives you a manual. A partner works with you to understand your specific workflows, co-creates a robust implementation plan, and provides ongoing support that goes beyond a standard help desk.

Ultimately, the most valuable AI agent is one that is backed by a partner who is in it for the long term, not just for a quick sale.

Photo: Yuichiro Chino, Getty Images

This message appears via the MedCity Influencers program. Anyone can post their views on healthcare business and innovation on MedCity News through MedCity Influencers. Click here to find out how.