Survey: Where AI Reduces Toil and Where It Still Falls Short

We’re so glad you’re here. You can expect all the best TNS content to arrive Monday through Friday to keep you on top of the news and at the top of your game. Check your inbox for a confirmation email where you can adjust your preferences and even join additional groups. Follow TNS on your favorite social media networks. Become a TNS follower on LinkedIn. Check out the latest featured and trending stories while you wait for your first TNS newsletter. DevOps teams spend nearly a third of their time on manual work, according to a DevOps report from DuploCloud. If you have 50 engineers making an average salary of $200,000, that translates into $3 million wasted each year, assuming that all of the manual work can be automated.A new survey of 135 engineering and DevOps leaders across North America and Europe reveals where AI is helping teams reclaim time and where it still struggles. The respondents come from startups and mid-market Software as a Service companies. This includes technical founders, platform engineers and compliance leads.The report found a 67% increase in AI investment last year, which can address priorities like reducing manual toil. But the results are mixed.AI does improve some targeted workflows. At the same time, it often adds noise and integration overhead.Here’s a look at the survey findings combined with industry research. With this data, you can clarify where AI delivers measurable value and where the technology still falls short.The point of DevOps is to accelerate software delivery. And yet it still manages to slow teams down. Thirty percent of engineers’ time still goes toward manual, repetitive infrastructure tasks, audits and tool maintenance.We’ve found that the biggest pain points include:One platform lead noted, “Everything feels duct-taped together. One script breaks, and it’s all hands on deck.”The result is slower releases, mounting audit fatigue and burnout. Almost half of the respondents report that their exhaustion is directly tied to DevOps workloads. This echoes Google’s 2024 “DORA Report” that links unstable priorities to lower productivity and higher stress.Of course, teams aren’t waiting for ideal conditions to get started with automating. Sixty-seven percent of those surveyed said they’ve increased AI investment in the past year.Here are their stated top goals:As one senior engineer put it, “Half my job is copy-pasting scripts and hoping nothing breaks.” For many leaders, AI isn’t an experiment anymore. It’s a matter of survival.Speed has become the north star. Fifty-four percent of teams ranked deployment frequency and lead time to change as top priorities.This signals a clear shift toward faster, more reliable delivery.2025 marked the rise of AI agents. These are systems that observe, decide and act in real time.These agents can execute infrastructure tasks safely through built-in controls, like:Nearly 80% of leaders said they’re open to using AI agents. But they need the tools to integrate well and maintain transparency.Of course, context will continue to be the difference between helpful and harmful AI.Without access to an organization’s real, live infrastructure data, AI systems lack the much-needed context and simply can’t act effectively. In fact, the survey notes that “without infrastructure context, responses can be vague and inactionable.” Teams that connect AI to live systems, from IaC and CI/CD to observability, see better accuracy and faster resolution.AI correlation across logs and metrics can reduce mean time to resolution (MTTR) by 20% to 40% in well-instrumented environments. It identifies related incidents and recommends proven runbooks, saving engineers from manual searches.Teams with inconsistent telemetry typically see smaller gains, usually maxing out at 25% reduction of MTTR.Clean, labeled data remains the key constraint.Sixty-two percent of respondents said the complexity of compliance and security is one of their top operational burdens. Automated evidence collection can map infrastructure to frameworks. These include SOC 2. Plus, they can generate audit-ready reports. This often cuts preparation from weeks to days.Auditors still require human review for access controls and data-handling practices. So automation simplifies proof gathering but does not replace attestation.Drift detection remains one of the most reliable AI use cases. Teams that adopted automated configuration-comparison tools reported finding drift within minutes. It’s much better than doing daily manual checks.Why it works:This use case consistently delivers measurable returns and rapid adoption.Anomaly-detection systems often produce 15% to 25% false positives during early deployment. Each one requires manual triage and slowly undermines confidence.How to limit false positives:Without these guardrails, teams revert to ignoring alerts altogether.More than half of teams target cost reduction through AI, yet recommendations frequently miss the business context. Algorithms may suggest deleting “idle” disaster-recovery resources or downsizing burst-capacity services.AI can highlight trends but cannot fully understand risk, redundancy or seasonality. Pair every cost recommendation with human validation to avoid unplanned outages or compliance gaps.The survey found that most organizations are still script-driven. That, or they’re partially automated. AI tends to magnify your existing inefficiencies in these environments.If you can get your team to a higher level of DevOps maturity with proactive monitoring and Policy as Code foundations, you’ll see the strongest ROI.The evidence is in, and engineering leaders are voting with their budgets.Meanwhile, DevOps engineers are moving from manual execution to automation architecture and AI-assisted operations.One CTO summarized it thus: “We’re not replacing people. We’re upgrading their impact.”The survey highlights four consistent habits among top-performing teams:Teams that integrate AI directly into CI/CD, policy engines and observability systems report the highest efficiency gains. They treat AI as an execution layer, not just a recommendation engine.AI is changing DevOps incrementally, not radically. Reducing manual work from 30% to 20% is a realistic and valuable goal, roughly equivalent to adding one engineer for every 10 on staff without increasing headcount.As one leader noted, “If you’re not using AI for DevOps yet, you’re already falling behind.”Teams that achieve success share four habits:AI’s greatest value in DevOps lies in amplification, helping people focus on design, reliability and judgment rather than repetitive tasks.The insights we’ve included here come from our “State of AI in DevOps 2026 Report,” published by DuploCloud.Read the full report: “State of AI in DevOps 2026.” It reveals a field shifting from scripts to agents and from manual firefighting to automation-first operations.In short, teams need to combine automation maturity, contextual AI and strong governance. That way, you’ll build faster, operate more safely and scale with confidence. Community created roadmaps, articles, resources and journeys for developers to help you choose your path and grow in your career.