5 Essential Questions to Consider Before Implementing Agentic AI
Understanding Agentic AI in Healthcare
Agentic AI, which refers to artificial intelligence systems capable of making decisions and taking actions independently, is making waves in healthcare. A prime example can be seen at NYU Langone Health, where AI is actively utilized to enhance the education of medical students.
How Agentic AI Works
At NYU Grossman School of Medicine and NYU Grossman Long Island School of Medicine, an AI application integrates data from over six inpatient hospitals and 300 outpatient locations. This system matches electronic health records with important research and diagnostic information, sending findings to students even before they start their day.
According to David Brauchler, a technical director at NCC Group, the goal isn’t just to improve the AI itself, but to make surrounding systems more effective, which enables AI to deliver better, more relevant decisions.
Key Questions for Implementing Agentic AI
Before integrating agentic AI into a company’s operations, it’s crucial to address several important questions to ensure effective deployment.
1. What Problem Will Agentic AI Solve?
Companies must start by identifying the specific problem that agentic AI will address. Developing a strategic plan is essential to ensure that AI applications are seamlessly integrated into workflows. Nader Mherabi, EVP and Chief Digital and Information Officer at NYU Langone Health, emphasizes that if AI operates outside of existing processes, it will likely be underutilized.
By targeting straightforward use cases, such as matching medical cases to relevant research, organizations can achieve significant improvements. Mherabi notes that the AI system doesn’t interact with sensitive patient data, making it a low-risk solution.
2. Do We Have the Data to Make It Work?
The success of an agentic AI system heavily relies on the quality and relevance of the data available. Tom Coshow, a senior director analyst at Gartner, explains that the data must be well-aligned with specific use cases. Ambiguous or poorly organized data can lead to incorrect insights, or worse, unauthorized access to sensitive information.
Maintaining data integrity is another critical aspect, necessitating skilled personnel who specialize in data management for AI applications.
3. Will Data be Secure?
As organizations adopt agentic AI, they must also consider security risks. Each addition to a technological ecosystem may expose the system to new vulnerabilities, according to Brauchler. Companies should conduct thorough security assessments when integrating AI technologies.
Testing can reveal potential risks, as seen in a recent evaluation by NCC Group, where the integration of AI inadvertently created security gaps in existing platforms.
4. Where Does Agentic AI Fit into the Company’s Tech Strategy?
Understanding how agentic AI complements overall technology strategies is vital. CIOs should evaluate whether it aligns with their current systems or if it requires partnerships with specialized vendors. Additionally, it’s important to determine how many agentic AI applications the company is ready to manage and monitor effectively.
New York University takes a modular approach, developing small, manageable AI agents that work together seamlessly. This method allows for individual components to be adjusted as needed without disrupting the entire AI operation.
5. How Will Agentic AI Be Monitored?
Implementing agentic AI is not a one-time effort; continuous monitoring is necessary to ensure the AI is functioning as intended, using data appropriately, and providing real benefits. Mherabi clarifies that ongoing attention is crucial for the success of these applications.
At NYU Langone, consistent oversight has enabled their AI platform to remain effective since its launch. It was designed to deliver tailored educational content to students, and its success in achieving this goal demonstrates the importance of sustained monitoring and support.
Agentic AI has the potential to revolutionize medical education and many areas of healthcare. By asking the right questions and ensuring robust systems are in place, organizations can harness its capabilities effectively.
