Talk to the Data: AI Agents and the Decline of Traditional Practice Software

Written By Alan N. Glazier, OD, FAAO, DIPL. ABO
With editorial contributions by Sweta Patel, M.Optom, MS.Health IST-MP.

“Software is dead” is a common phrase you’ll hear among the younger generation of software developers. What this indicates is that with the advent of AI, ultimately, software won’t be needed to run most applications, including those in our offices. In other words, software will become obsolete, not because code disappears, but because more adaptive, AI-driven systems are replacing traditional, static software. In order to understand why, it’s important to first understand what software really is and does, and what replaces it.

For simplicity, think of traditional software as a middleman – a layer of technology that exists between a user and the user’s data. The data typically exists in a database, and the software is programmed with logic that makes sense of the data it interacts with, processes it, structures it, and produces a useful output that the user engages with. For instance, your EHR software takes patient data and puts it in a useful and meaningful form for you to manage patients with—the interface you enter information on for every patient, every day. Practice management software takes patient demographic data and structures it to generate practice performance data, invoices, manage ARs, and other functions. Customer communication software takes patient data and uses it to generate emails, recalls, text reminders, create newsletters, and run a host of other applications. QuickBooks software takes data and structures it into meaningful insights on the financial aspects of your practice.

While software has greatly simplified our practices, it comes with several downsides. For instance, it is expensive—not only the cost of the software itself, but also annual contracts and update fees. Traditional software is severely limited in terms of customization, often fails to interact with other software, and these gaps come at the cost of efficiency. Software is usually specialized, so you need different software for different applications; no one software does it all. While newer modular and API-driven platforms are starting to address some of these limitations, the reality in most practices today is still a fragmented and costly software environment.

Forward to the era of artificial intelligence (AI). AI systems can take data, process it into useful information, structure it, and produce an output that the user can engage with. Unlike traditional software, dynamic AI software can learn from data and apply what it has learned to improve functionality, making them more agile and adaptable than traditional software. AI can identify mistakes and inconsistencies in data, correct them, and even assist in writing code to perform the necessary tasks, thereby reducing the need for developers to manually rewrite software. While AI isn’t yet capable of fully self-writing, testing, and deploying production-level healthcare software safely, its ability to adapt and automate continues to evolve rapidly.

At some point in the not-too-distant future, AI will transform the traditional software layer into a more intelligent, adaptive system, processing all information and performing logic functions that were once hardcoded into software. Instead of relying on software to run a practice, it will be the database and the data that matter most. AI will enable more direct, dynamic engagement with practice data, reducing reliance on rigid application layers. Rather than choosing the best software for your practice, you’ll choose the best database for your needs—and you’ll do this by engaging with the AI through an agent: chatbot with a simple interface, similar to ChatGPT, Claude, or other conversational tools you already use.

AI agents automate tasks that traditional software can’t handle. For example, AI is capable of processing unstructured data, such as images, videos, and language, in ways that traditional algorithms cannot. This allows AI agents to perform tasks such as medical diagnosis, generation of pricing, or creating easy-to-understand personalized patient education and management plans. As agents learn, they become more efficient, unlike software, which remains static, thereby further increasing these efficiencies.

AI agents can handle multiple functions, thereby reducing the need for disparate software systems to communicate or integrate to achieve complex tasks. This reduces redundancy, simplifies maintenance, and lowers costs. Another advantage of AI agents is their ability to write code to solve problems and generate desired outcomes. Think about it as talking directly to your data through a lightweight, intelligent interface—one that understands context and delivers outputs more powerful than those produced through traditional software.

There are several constraints that need to be addressed before AI significantly impacts healthcare software. First and foremost are HIPAA concerns. Next are concerns about unauthorized access, data breaches, or improper handling of sensitive health information. Secondly, AI systems need to be “trained” to produce meaningful, useful and accurate information. Until properly trained, they can produce incomplete and incorrect results, which have obvious negative implications when it comes to potential misdiagnoses and basing treatment and management decisions for patients solely on AI outputs.

By minimizing reliance on rigid software layers and enabling AI to interface more directly with data, physicians can query information and execute practice management and patient care tasks more naturally and intuitively. However, this transformation doesn’t come without challenges. There are ethical concerns, data privacy issues, and a risk of over-dependence on AI. Despite these challenges, the potential of AI agents to redefine traditional software structures is promising. As AI continues to mature, it is likely to reshape how humans interact with technology, making AI agents a cornerstone of future medical practice.

Follow Dr. Alan Glazier on LinkedIN and join the ODs on Technology community for updates, insights and breakthroughs in Optometry