What is agentic AI and what does it mean for healthcare?

Artificial intelligence is advancing every day, bringing new capabilities — and new buzzwords — to the healthcare industry at a breakneck pace.

The latest tech term to catch fire is “agentic AI” – a new twist on how artificial intelligence models can bring increased efficiency to healthcare processes, from the administrative environment to the bedside and beyond.

But what exactly is agentic AI, and how will it revolutionize the way healthcare organizations operate?  Let’s take a look at some of the possibilities. 

What is agentic AI? 

Agentic AI is when artificial intelligence is designed to act with human-like autonomy to complete a specific task. Also known as an “AI agent,” this type of AI doesn’t require direct human oversight to meet goals by solving problems, making decisions, and learning from previous results to adapt in the future. 

However, not all AI agents are agentic AI.

AI agents exist on a spectrum. Some are relatively simple, like an RPA bot that automates billing tasks based on predefined rules. These systems can efficiently follow instructions but may struggle with unexpected scenarios and complex tasks. In contrast, agentic AI represents a more advanced form of AI agent. It’s designed to operate with greater independence, adapting dynamically to changing conditions. For example, an agentic AI staffing tool could proactively adjust nurse assignments in real time to account for sudden absences or unexpected patient surges — all without human intervention.

Consider the example of autonomous vehicles. A driverless car is designed with an overarching goal in mind: get a passenger from A to B without crashing. Once it’s turned on, the car becomes its own master. Using complex AI in the background, the vehicle identifies traffic signals, avoids obstacles, navigates roads, and parks at its destination without human intervention (unless it encounters a problem that exceeds the limits of its programming). 

Other examples may include:  

• AI customer service agents that can make recommendations, answer questions, or direct consumers to appropriate resources 

• Cybersecurity sentinels that constantly monitor networks for intrusions and send alerts or scramble defenses accordingly 

• Human resources assistants that scan resumes for keywords, prioritize candidates based on defined criteria, schedule interviews with prospective employees, and communicate with applicants 

• Unmanned drones that deliver goods, perform surveillance, or assist with military and security operations 

• Tools for monitoring and managing physical systems, such as water and gas networks, manufacturing processes, traffic systems, or public transport networks 

Agentic AI typically leverages a combination of established AI strategies, including large language models (LLMs), natural language processing (NLP), computer vision, machine learning, deep learning, neural networking, and automation technologies.  Bringing all of these approaches together in the right way can create tools that require little in the way of active intervention by their human creators. 

How does agentic AI differ from generative AI?

Agentic AI is built for completing tasks. It has its instructions, and it executes on those instructions according to its own idea of best practices for getting to the right answers.  

In contrast, generative AI is focused on creating new content based on what it knows about its training data. Its strength is in identifying patterns in massive datasets and using those patterns as the basis for developing output in the form of creative or summary text, videos, audio content, or images. It typically requires a prompt, originating from a human source, to provide instructions about what content to create.  

That doesn’t mean generative AI and agentic AI are totally separate categories of tools, however. AI agents can harness generative AI capabilities to complete their assigned tasks, because sometimes the agent needs to generate something new as a step toward solving a problem or meeting an objective. 

For example, an AI agent designed to triage a patient’s clinical question to the right provider may need to use generative AI capabilities to scan through and summarize the patient’s medical records and utilization history so it can choose an active member of the care team to whom to direct the inquiry. 

In other words, agentic AI can automatically and autonomously act on the output of generative AI models in the same way that a human would take the results of a generative AI prompt and use it to complete a task. 

How will agentic AI impact healthcare? 

Healthcare industry leaders are very excited about the possibilities for agentic AI to improve efficiency and reduce complexity across the care continuum. There are dozens of examples of how agentic AI could support healthcare applications, such as: 

• Optimizing staff scheduling and managing patient flow to ensure the availability of space and resources at all times 

• Automating repetitive, time-consuming revenue cycle management tasks, such as completing prior authorizations or submitting claims 

• Assisting patients with scheduling appointments, meeting financial responsibility, renewing prescriptions, and accessing medical records 

• Providing care navigation services by completing health plan eligibility checks, recommending appropriate care providers, and helping with prescription drug access 

• Ensuring adherence to clinical best practices by assisting with treatment planning, clinical decision support, and follow-up tasks 

• Improving supply chain efficiency by predicting resource utilization patterns, automatically reordering supplies accordingly, and managing vendor relationships 

• Strengthening cybersecurity by providing continuous monitoring of network activity, quickly flagging potential vulnerabilities, and deploying defenses against unauthorized access 

• Augmenting coaching, education, and remote patient monitoring resources for chronic disease management, hospital at home programs, behavioral healthcare services, and other high-needs clinical areas 

Some of these use cases are already being put into action, albeit with varying degrees of true autonomy on the part of AI systems. Few healthcare organizations currently have the digital maturity to truly let agentic AI off the leash just yet, but it’s only a matter of time before AI agents play a much larger role in everyday healthcare system operations. 

Fully agentic AI is a tantalizing prospect, but it also comes with many unanswered questions around safety, reliability, and bias – not to mention the social, economic, and ethical challenges of theoretically replacing human staff with autonomous algorithms. 

With regulation and best practices for AI deployment still in their infancy, healthcare organizations will need to pay close attention to ensuring that enthusiasm for agentic AI doesn’t outpace the guardrails that will keep patients – and employees – safe from unintended consequences. 

Nevertheless, the advent of agentic AI is exciting and represents the next real step toward reducing avoidable burdens on the healthcare system. If deployed carefully and intentionally with an eye toward improving experiences instead of just slashing costs, agentic AI could dramatically revolutionize the way humans and machines collaborate to deliver more efficient and effective healthcare services.  

Jennifer Bresnick is a journalist and freelance content creator with a decade of experience in the health IT industry.  Her work has focused on leveraging innovative technology tools to create value, improve health equity, and achieve the promises of the learning health system.  She can be reached at [email protected].