5 ways to use AI to modernize your legacy systems

Follow ZDNET: Add us as a preferred source on Google.

ZDNET’s key takeaways

• Managing technical debt can consume up to 40% of IT development time.
• One way to overcome the legacy challenge is to use specialist AI agents.
• Focus on testing tools, refining projects, and pushing long-term change.

Businesses are being held back by their legacy systems. Research from IDC reports that unmanaged technical debt can consume between 20% and 40% of IT development time, diverting resources away from innovation and modernization.

IDC suggests many enterprises are eager to deploy AI-enabled services, but their ambitions are constrained by technical debt, including outdated systems, fragile integrations, and limited data interoperability.

Also: 5 ways to grow your business with AI – without sidelining your people

The good news is that pioneering executives are meeting this challenge head-on. While the legacy IT burden can prevent organizations from deploying new data and AI services, some businesses are taking a radical approach — they’re using AI to modernize their systems and create new opportunities for internal development teams.

That’s certainly the case with Jeff Love, CTO at the Professional Rodeo Cowboys Association (PRCA), a governing body for the sport that sanctions rodeo events in the United States, Canada, and Mexico.

Love was eager to explore how AI could help his organization, which has close to 100 years of history, overcome its intractable legacy IT challenge. Here, he suggests five lessons for other business leaders who want to take a similar approach.

1. Test AI models

Love explained to ZDNET how a large portion of PRCA’s backend systems ran on AS/400 code that’s 40 years old.

This reliance on legacy systems meant the development team spent more time maintaining old code than building new capabilities, which prevented the organization from embracing digitalization and new ways of working.

“That’s been my goal here — to modernize our applications, just because they’re becoming difficult to maintain, and there’s a lot of knowledge that we’re losing about how to maintain these systems as they get older,” he said.

Also: 5 ways rules and regulations can help guide your AI innovation

Love recognized that AI might offer a way to help PRCA overcome its legacy challenge. However, initial tests with gen AI models a year ago produced mixed results.

“I tried using ChatGPT, but the difficulty it faced was just with the amount of code. ChatGPT couldn’t handle the amount of data that we tried to give it. There are probably close to 1,000 files that it was trying to summarize,” he said.

“Then, from reading up on Grok, I thought it might handle some of the code a bit better. But I tried, and it just couldn’t do it. I tried a few other tools that claimed they could document your code base. However, they didn’t look at all the files holistically. They would look at each file and document that data.”

2. Use a specialist solution

After his initial explorations with gen AI models, Love began working last July with Zencoder, an agentic platform that analyzes business logic and translates it into plain-English explanations.

For an organization eager to reduce its technical debt, Love said the platform sounded like a dream come true.

Also: 5 ways Lenovo’s AI strategy can deliver real results for you too

While traditional approaches and gen AI models failed to penetrate decades of business logic, he believed Zencoder could help PRCA overcome its legacy code challenge.

“I tried Zencoder out,” he said. “I gave it our AS/400 code, and I said, ‘Document this and give me the business rules, give me what database files are accessed, and explain how we could modernize and what we should be taking into consideration.'”

Love said the initial results were promising, but not faultless: “It wasn’t perfect at first, just because of the sheer amount of information it had to go through.”

However, his team refined the work of the agents, and business analysts at PRCA recognized they had a tool that could help shift the organization away from its AS/400 systems: “That’s where we started creating more detailed requirements.”

3. Put theory into practice

Love and his colleagues gave the agents instructions, guidelines, diagrams, and workflows. These key requirements helped produce a wiki for PRCA’s business analysts.

The organization then created wireframes that drew on the key requirements and business rules.

“Based on those wireframes, I’ve been able to take the work items, put them into an agent that I’ve created to help us with coding, and then it takes the workflow, puts it into the UI structure that we’re utilizing in the modernization, and then we use that as our starting point for coding,” he said.

Also: 5 ways to prevent your AI strategy from going bust

Love said the Zencoder technology helped staff understand the interconnected nature of code and systems.

Then, as the platform generated new code and modernized legacy systems, it created unit tests that prevented bugs before production.

“We were able to put in our requirements, put in test acceptance criteria to make sure that we were capturing the business rules, so that when we did modernize the system, that we were still taking into account the actual rules,” he said.

4. Shift from legacy thinking

Today, PRCA’s technology team is embracing modernization. With the help of AI, time previously spent on legacy concerns is directed to digitalization.

Love estimated that Zencoder supports a 50% reduction in development times, which the IT team uses to build digital services, create new event management tools, and deliver better experiences.

“We don’t have lots of resources,” he said. “Our small internal team has got a half dozen large systems that we’re in charge of managing, which becomes overwhelming at times for all the support work that’s needed to keep functioning.”

Love said rodeo’s complex business logic means it can take new hires a long time to learn the rules of the sport.

Also: This company’s AI success was built on 5 essential steps – see how they work for you

Zencoder takes the hard work out of the process, allowing staff to get up to speed quickly and to focus on changes that deliver the most value to the organization.

“We can bring a developer in, and they can work on day one. They have a better understanding of the logic, so they’re not as afraid of making changes, because they’re able to see what the actual business rules are,” he said.

“Now we spend more time on things like unit testing, all of those things that are crucial to building a solid application, but that, unfortunately, can sit on the back burner because you’re so focused on functionality and getting a product out.”

5. Find new challenges

Love said his team aims to complete the organization’s AS/400 system migration by the end of 2026.

Once that work is complete, the team will tackle the next legacy platform, PRCA’s ASP.NET Web Forms technology.

“The first goal is that we’re going to bring ourselves up to date,” he said. “We’re working 40 years in the past. Once we get off the AS/400, we’re 20 years in the past. The next big project will be migrating off ASP.NET to a more modern application.”

Love said the long-term aim is that the team’s agentic-enabled processes will help the organization continue evolving digitally.

“I’m two years into my five-year plan, which is to bring us into modern times,” he said. “But then, at the end of that period, there will be other projects that we started at the beginning, and it’ll be time to go in, revamp them, and handle new business rules.”