The difference between the artificial intelligence of today and what agencies used five or even three years ago can be summed up in one word: brittle.
The latest AI tools and capabilities are more dynamic and flexible than earlier versions such as robotics process automation (RPA), said Ryan Macaleer, vice president of software for federal at IBM.
AI tools, particularly generative AI capabilities, have a more fluid nature now, he said during Federal News Network’s AI and Data Exchange.
“They can respond in real time to inquiries, to forms, to understanding at scale versus the way in which you had to deploy an RPA bot, which was based on a particular interface, a particular screen, a particular back end. When that screen changed or that back-end service changed, the bot could break down and interrupt that process,” Macaleer said.
“There’s just so much more application for these generative AI solutions. A lot of times when we think about RPA, we’re thinking about bots. We’re thinking there’s this easy way to align chatbots and RPA bots because of the similar nature there. But AI, particularly generative AI, is so much more than chatbots.”
Using AI to reduce inconsistencies, eliminate errors
GenAI, for example, can automate mundane processes and give users more insight into the unstructured and structured data.
Macaleer said IBM works with one agency that has more than a thousand documents that it needs to process to do claims and benefit processing.
“As you can imagine, that’s a process that is error prone, and even with something like RPA, the complexity of having to manage all those documents and the scripting and the rigidity there makes it really, really difficult,” he said. “Through our generative AI approach, we’ve been able to really reduce those inconsistencies. We’ve been able to eliminate those errors, help expedite that decision-making at the time in which it’s going through that process and deliver a much, much better outcome for both the citizen and the employee.”
Macaleer said it’s the outcomes discussion that really drives the use and benefits of AI. That is especially true when agencies start applying these tools to IT modernization and optimization initiatives.
Applying AI to legacy code
One growing AI use case for agencies is modernizing legacy code.
Macaleer said IBM is applying code modernization AI tools internally to gain efficiencies.
He said AI can help with things like modernizing the IT stack; pushing software updates, patch management, security fixes; and improving the efficiencies of code generation.
“It’s really bridging the gap between some of those older technologies and newer technologies, improving the data interoperability, reducing the time that humans are spending on those system updates,” Macaleer said.
“With RPA, for example, they could just never have really delivered on this modernization story. The right application migration strategy can ensure that the right applications are chosen and are not only compatible with their modern technology but also are adaptable to fit into their legacy systems. It’s a unique differentiator of this new AI paradigm.”
The Office of Personnel Management recently received $18.3 million from the Technology Modernization Fund to use AI to modernize COBOL code.
Macaleer said the OPM example is one of many happening across government. Although some agencies might be in the early stages of using AI for code development, there is a lot of excitement.
“It’s really about striking the right balance between what are those workloads that are the most important to the enterprise — the most needed for security, the most needed for performance and uptime — and what are those workloads that are still running on a platform that doesn’t fit that criteria and can be migrated or can be upgraded to something different,” he said. “That balance is something that we engage in conversation about all the time, and we do have tools to help in that journey.”
Integrating AI into FinOps
Another area where AI is showing promise is in FinOps, which helps organizations apply cost optimization principles to their cloud environments.
Macaleer said the idea is to use Technology Business Management and other methodologies to get clarity on where money is going and whether it is delivering value to the agency.
“One of the most effective or most beneficial things that we’ve seen from our government clients is being able to, in real time, understand what applications are being called on, the resource demand for those applications and then create a resource pool that can serve up those resources in real time, instead of having a static resource pool dedicated to each application,” Macaleer said.
“What that does is allow you to optimize all those resources — the compute, the storage, the memory — and reduce the cost that you’re spending on those resources while actually providing a more performant application. You’re driving down the cost in real time, automating those actions and then providing a better outcome to the user of those applications in terms of that application performance.”
Discover more articles and videos now on our AI & Data Exchange event page.
Copyright
© 2025 Federal News Network. All rights reserved. This website is not intended for users located within the European Economic Area.
