Artificial intelligence (AI) is helping IT service management (ITSM) teams save time, but it is not always making the work lighter. New SolarWinds research, based on an online survey of 844 qualified IT professionals across North America, EMEA, and APAC, found that respondents were saving time on issue detection, end-user support, and ticket triage. Eighty-four percent said AI had met or exceeded their ROI expectations.
At the same time, the survey shows why task-level efficiency does not automatically reduce total workload. Fifty-two percent said their team’s workload increased after adopting AI, while 19% said it stayed about the same. Only 30% reported a decrease. These gains coexist with new responsibilities, including maintaining integrations, validating outputs, improving data, training staff, and managing exceptions.
AI in ITSM is delivering measurable gains
The report found that 84% of respondents said AI had met or exceeded their ROI expectations. After an average of roughly 16 months using AI in their ITSM environments, respondents reported average weekly time savings of:
- 3.3 hours detecting and flagging issues
- 3.0 hours responding to end-user requests
- 2.9 hours triaging tickets
But the reported impact on organizations extends beyond task-level efficiency. Respondents also reported improvements in areas including employee productivity, Service-level agreement (SLA) compliance, mean time to detect, and mean time to resolve (MTTR).
In the previous State of ITSM research on generative AI features in SolarWinds Service Desk, average resolution time fell from 27.42 hours before enablement to 22.55 hours after enablement, a 17.8% reduction. In the 2026 survey, 86% of respondents said AI improved employee productivity, including 47% who described the improvement as significant.
These results matter because they show that while AI adds work for technicians, it also reduces repetitive work, speeds incident response, and makes it easier for employees to get support. The survey measured task-level efficiency and total workload, not whether IT service desks felt simpler to operate or whether service desk agents experienced less day-to-day strain.
Why AI can add to ITSM work levels
The headline contradiction is straightforward: 52% of respondents said their overall workload increased after adopting AI. The finding will be familiar to teams already responsible for running an IT service desk.
Among the 428 respondents who said AI had increased their team’s workload, 48% cited managing and maintaining AI tools and integrations as added work. Forty-seven percent cited reviewing and validating AI-generated outputs, and 37% cited training and fine-tuning AI models.
| Among respondents reporting increased workload, selected sources of added work | Share |
| Managing and maintaining AI tools and integrations | 48% |
| Reviewing and validating AI-generated outputs | 47% |
| Training and fine-tuning AI models | 37% |
| Onboarding and training staff on AI tools | 35% |
| Handling errors or failures from AI automation | 33% |
| Managing vendor relationships and contracts | 29% |
| Governing and auditing AI decisions and actions | 27% |
| Documenting and updating processes | 11% |
*Question allowed respondents to select up to three options; percentages are based on 428 respondents who reported increased workload.
Essentially, though a faster triage process is valuable, it doesn’t erase the effort required to connect AI to the service-management environment, establish guardrails, identify poor answers, and keep the workflow from breaking when data or processes change.
Sean Sebring, Solutions Engineer Manager for IT Service Management at SolarWinds, described the workload issue as part of a new capability organizations are still learning to manage. “There’s a lot that needs to be learned when it comes to the management and integration of AI,” he says. “This perception that AI requires work will likely change as organizations become more aware and comfortable with AI, and more users are trained on the technology.”
But how can IT service desk teams overcome this? Sebring continues, “There are more and more foundational best practices being made available, such as ITIL 5, and more standard training emerging around integrating and implementing AI available to help technicians adapt to AI.”
The hidden cost of AI adoption is ongoing work
Cost is another area where AI adoption looks different once teams get past the initial business case.
Only 7% of respondents said they encountered no unexpected costs from AI adoption.
The most common unexpected costs were staff training and upskilling (48%), data quality and cleanup (47%), and ongoing tuning and maintenance (45%). These are recurring parts of the operating model and should be included in annual planning and budgeting.
The survey found that 83% of respondents spend at least three hours each week on AI monitoring and maintenance, and 44% spend more than six hours.
This further illustrates that the cost of an AI initiative extends beyond the software, model access, or initial implementation. It includes the time spent maintaining integrations, improving inputs, reviewing output, training teams, and adapting processes as the organization learns where the technology helps and where it does not.
For ITSM leaders, this is where a narrowly framed efficiency case can fall short. If success is measured solely on the number of tickets processed or automated interactions completed, the organization may miss the additional work shifted to service desk agents, service owners, platform teams, and administrators.
Want an accurate measure of productivity? Measure outcomes, not activity
The report points to a revealing difference in how organizations assess AI. Only 21% of respondents described their measurement approach as outcome- or experience-oriented. In this survey, organizations using primarily activity-oriented measures were 2.4 times more likely to report increased workload than organizations using outcome-oriented measures.
Activity metrics aren’t useless. Ticket deflection, triage volume, response time, automation completion rate, and knowledge-article creation can all be useful indicators. The problem comes when those measurements become the whole story a team’s AI adoption is based on, ignoring the time service desk agents spend validating AI suggestions instead of processing tickets. AI can help close incidents faster, but unresolved root causes keep generating repeat demand, and publishing more knowledge content doesn’t help when employees still struggle to find accurate answers.
Outcome measures are harder because they ask whether service is improving over time. Depending on the organization and use case, this might include:
- Mean time to restore service
- First-contact resolution
- Reopen and repeat-contact rates
- SLA performance for priority incidents
- Employee effort required to obtain support
- Agent experience and the quality of work
- Reduction in recurring incidents or avoidable demand.
Sebring notes that the research did not fully examine why respondents perceived that AI had not reduced workloads: “The report findings make it very evident that AI makes organizations faster and more efficient. But the report did not deeply analyze why people perceived that AI did not reduce workloads.” He continues, “In the future, we may focus on more experience-centric metrics, looking at whether servce desk agents find their work more meaningful, or whether the quality of the service has improved.”
For ITSM, faster is not always better if faster simply means a human has to spend more time correcting, checking, or routing work that should’ve been automated. The desired outcome is not AI activity. It is better service, less avoidable work, and a healthier operating environment for the people responsible for delivering it.
Data and integration still decide the outcome
It’s easy to treat AI as the maturity layer that sits on top of ITSM. After all, if you’re quickly adopting AI, you must be mature, correct? The reality is no. In practice, AI exposes the maturity gaps existing beneath it.
Because AI depends on the quality of the data, knowledge, workflows, and ownership structures available to it, gaps in categorization, knowledge, tooling, configuration information, or process ownership can reduce the reliability of AI-assisted work.
Sebring put it plainly: “There are two key factors that make the difference between success and failure when implementing AI in ITSM: the data foundation and integration complexity. Organizations must have a good data foundation for AI to leverage, and they must have AI tools that seamlessly integrate with their toolsets.”
The report supports that emphasis. Data quality and cleanup were the second most frequently cited source of unexpected cost, reported by 47% of respondents. Disconnected tools also create more work, because every additional handoff or integration point becomes another component to maintain, monitor, and troubleshoot.
This means that data work should be treated as part of the AI strategy rather than a cleanup project that can be deferred indefinitely.
Optimize your IT service desk before you automate
For teams deciding where to begin, the most useful advice may be the least flashy.
“The ITIL guiding principle ‘start where you are’ ensures that you get systems and data ready to be leveraged by AI,” Sebring said. “Organizations should optimize and then automate – specifically in that order.”
Before adding AI to a workflow, ask whether the workflow itself is clear, repeatable, and worth scaling. A solid starting point is a high-volume, well-defined task where the organization can measure results and quickly identify issues. Ticket triage, issue detection, incident documentation, and routine request support are all candidates when the necessary data and feedback loops exist.
The point is to make the first use cases manageable enough that teams can learn, measure, refine, and expand without creating an oversized maintenance burden.
ITSM work is changing, not disappearing
The research does not show that AI has failed to deliver value in ITSM. Respondents reported productivity gains and continued investment in AI, even as many teams took on additional work to maintain, validate, govern, and improve these systems.
For many teams, AI is replacing some repetitive tasks with new responsibilities: maintaining systems, validating results, improving data, managing exceptions, and redesigning workflows around the technology.
The lasting value comes when organizations use the time AI saves to strengthen the underlying service-management system, rather than allowing it to be absorbed by unplanned maintenance, disconnected tools, and avoidable rework.
AI is not simply removing work from the IT service desk. It is changing the type of work teams do. The clearest path forward is to identify and improve the work that needs attention before automating it.
* The 2026 State of ITSM Report is based on an online survey of 844 qualified IT professionals across North America, EMEA, and APAC. Individual questions received different numbers of responses, so question-level sample sizes vary. The 2024 and 2025 comparisons cited here come from separate SolarWinds customer data analyses and should not be read as results from the 2026 survey.
Lauren Okruch
Lauren is a passionate Product Marketing Manager at SolarWinds, specializing in IT Service Management (ITSM). With a deep appreciation for the balance between structure and simplicity, Lauren focuses on creating solutions that reduce friction and improve efficiency. As an American living abroad, she brings a global perspective to her work and enjoys exploring the intersection of technology and human-centered resolutions.
