AI Use Cases in IT That Deserve to Go Mainstream
TL;DR
- Modernize legacy IT without breaking your bank: AI can prevent outdated business logic from creeping into modernized systems. Thoughtworks’ AI-powered tool, CodeConcise, analyzes legacy code, extracts business rules, and fast-tracks mainframe modernization. As of now, it has shaved 4 weeks off per-module transformation time and saved 240 FTE years in a large-scale overhaul.
- Boost developer productivity while reducing burnout: AI-driven IDEs, intelligent code suggestions, and automated workflows are minimizing cognitive load. Finastra’s AI-powered engineering pulse surveys identified an untracked 6-hour weekly workload spike. This insight helped leadership refine workflows, optimize task allocation, and reduce burnout risks.
- Keep product scope under control with AI-driven product intelligence: AI helps teams detect scope creep, automate sprint reporting, and predict under-scoped features before they disrupt development. CreateFuture used AI-powered backlog tracking to cut 30% of the time spent managing user stories and reclaim 4-8 hours per sprint.
- Enable self-healing IT to prevent costly downtime: AI is shifting IT operations from reactive monitoring to proactive, self-healing systems. Chamomile.ai’s LLM-powered log analysis detects subtle anomalies, predicts failures, and automates remediation. This has reduced manual investigation time and helped teams resolve issues before they escalate into major outages.
AI has made its way into just about every corner of IT and development. We hear about how it’ll boost engineering efficiency, improve work-life balance, and provide a smart, all-knowing, always-there assistant. It’s enough to make anyone’s head spin.
But despite being practically everywhere, AI is not quite delivering what we expected, or even anything at all. More than half of IT leaders are struggling to prove AI’s ROI and burning billions in the process.
In boardrooms, the focus is on agentic AI, risk, and the latest DeepSeek buzz, but there’s a bigger question no one seems to be asking: Are we actually solving the right problems with AI? Maybe the real challenge isn’t the technology itself but knowing where to apply it.
It’s not all doom and gloom, though. Some teams have figured this out. We sat down with a few AI trailblazers who are driving real impact with surprisingly simple (but effective) business applications. Here are the AI use cases they’re most confident in:
1. Modernize Legacy IT
AI is solving a long-standing challenge in legacy modernization: preventing outdated logic from creeping into modern systems. “Government agencies still operate on massive, monolithic systems,” says Noel Hara, VP & CTO public sector at NTT DATA. “Even when business logic is extracted or code refactored, it often includes outdated logic hidden deep in the code.”
Noel’s team is now using AI to analyze extracted rules and intelligently cross-reference them with current policies. Even better, the tools can suggest smart updates where policies have changed, all without overwhelming business owners. “This approach keeps the modernization process smooth and prevents a pile-up of technical debt in the new system.”
Thoughtworks has developed CodeConcise, an internal AI-powered accelerator that breaks down complex code structures, maps dependencies, and uncovers critical insights, even when key developers are long gone. The tool can also translate legacy code into modern languages to ensure compliance and alignment with the development team.
As of now, the internal rollout has been fairly successful: It has already slashed modernization time per module by 4 weeks and saved an eye-popping 240 FTE years on a full-scale mainframe transformation.
2. Improve Employee Productivity
Even before the current gold rush, AI was already handling the grunt work of generating boilerplate code and managing test cases– tasks that once drained hours from IT teams. Now, AI-led IDEs like Tabnine and JetBrains AI are layering these tasks with intelligent code reviews, contextual summaries, and automated knowledge documentation. Developers no longer have to juggle endless context switches or hunt for undocumented fixes.
Cycode’s field CTO Jimmy Xu can see the future of AI-assisted engineering panning out so much so that he terms AI as the “means” to enable 10x developers. He’s got a point. Step back, and AI starts looking less like a tool and more like the connective tissue of modern software development. It can stitch together fragmented workflows, run smarter handoffs, offer granular visibility, and surface inefficiencies teams might not even realize exist. Finastra experienced this firsthand after deploying a custom AI tool to monitor developer experience.
Initially, leadership used AI-driven pulse surveys to assess engineering well-being, but they soon uncovered an overlooked issue: developers were logging (previously untracked) six extra hours per week, thanks to relentless demands and continuous product updates.
Recognizing the risk of burnout, the leadership leveraged AI to optimize workloads, enforce focus hours, and refine communication hours to reduce unnecessary interruptions. Now, engineers could dedicate their energy to innovation without being bogged down by constant context-switching.
3. Manage Product Scope From “Creeping” Into Chaos
“For a software engineer, coding takes up only a small fraction of the workweek—about two hours a day,” comments Jeff Watkins, CTO at CreateFuture. Most of the week goes to tweaking user stories, epics, acceptance criteria, QA workflows, and, if you’re not careful, wrestling with scope creep that just keeps growing.
Fortunately, this is where AI has delivered us some of CreateFuture’s most impactful gains. “We’ve cut about 30% of the time spent writing, reviewing, and syncing up epics, stories, and acceptance criteria,” he explains, highlighting how AI has sharpened their product development flow.
His team now leans on AI to:
- Automate meeting summaries to track key discussions and scope deviations from the original scope and build an action checklist to course correct.
- Generate sprint reports by monitoring completed tasks, backlog shifts, and evolving requirements—helping teams spot scope creep before it derails progress.
- Run predictive analysis on past projects to catch under-scoped features early.
- Build feedback loops that summarize peer insights and surface growth opportunities.
Jeff sees AI as a ‘rubber duck’ for product scope: asking the right stuff, uncovering blockers, and tying engineering investments to business goals. “With AI handling sprint reports and doc reviews, we’re gaining back four to eight hours per sprint.”
4. Enable Self-Healing IT Infrastructure
A downtime is an expensive crisis for IT teams. Even on the good days, downtime can gobble as much as $9,000 per minute for large organizations. Factor in regulatory fines, data loss, and reputational damage for regulated industries, and the total impact can skyrocket beyond $5 million. Despite this, most IT monitoring tools remain reactive; they detect incidents but fail to prevent recurrence, and they struggle to scale. As Jimmy notes, "Scale and delivery become critical challenges in an increasingly complex environment."
A self-healing infrastructure bridges these gaps with autonomous monitoring tools. For Chamomile.ai’s CEO, Tirath Ramdas, log analysis is a prime example of how AI can ease developers' workloads while simultaneously strengthening autonomous IT systems. “Bug report analysis is tedious, rarely enjoyable, and often happens under high pressure when failures or outages are beginning to impact end-users,” says Tirath. Yet true root cause analysis requires more than just scanning for ERROR logs.
“It means correlating multiple logs and detecting subtle anomalies.” He uses LLMs to zap tedious log hunts and rapidly connect contextual patterns. “They’re great at fuzzy matching terms, intelligently diff-ing logs, and flagging unusual behaviors.” Even better, it can auto-detect small glitches in logs and cut down on manual investigations to accelerate incident resolution.
Beyond log analysis, AI tools can also auto-remediate minor incidents. Autonomous AI-driven systems continuously monitor real-time metrics—CPU usage, network latency, memory consumption—to detect anomalies, reduce false positives, and act when thresholds are breached.
If a system metric surpasses a predefined benchmark, AI can trigger an automated remediation protocol. “For remediation, AI can provide tailored remediation guidance and generate suggested code fixes with automation workflows. You can then use AI to quantify a normalized risk score for diverse findings and automate triaging,” comments Jimmy.
AI Can Drive Change if Your Culture is Built to Support it
AI can truly deliver time savings and bring full workflow visibility to engineering and IT teams– but only if the culture is ready to support it. “You don’t need fancy new tech to pull these results off,” Jeff says. It’s all doable with the AI tools already baked into your office suite, like Microsoft Copilot 365 or Google Gemini Advanced.
“What truly matters is investing in people. Show them how to get what they need from these tools,” he adds. Even in lean teams, spending a few hours coaching each person adds up and suddenly, you’re seeing a genuine return on investment.
Larger teams can try experimental projects like the infamous ‘hack the scope’ days. Throw your engineers an ambiguous project, a pile of data, and AI tools to solve problems like risk prediction or epic drafting. The best solution wins bragging rights, some incentives to encourage their next pilot project, and probably an over-the-top trophy. “So my takeaway is, get good at using the tools in front of you before you buy more, or risk becoming a victim of the old saying ‘all the gear, and no idea’."
Subscribe to the CTO Club’s newsletter for more insights on AI, business-engineering alignment, and self-healing IT.
