96% of IT pros now use AI: Top 7 agentic apps and key roadblocks

Nearly all IT and data professionals have adopted artificial intelligence in their workflows, but a deep dive into their actual usage reveals a more measured reality. A global survey of 700 data analysts and 700 IT leaders, conducted by Alteryx, shows that 96% of respondents now use AI, yet only half of them qualify as frequent users. Just 49% report using AI always or most of the time, while many still rely heavily on traditional tools like spreadsheets.

The push toward agentic AI is gaining momentum. Close to six in ten respondents, or 59%, anticipate actively deploying AI agents within the next year. Interestingly, at least half of those surveyed would grant these agents unrestricted access to their data, though the security implications remain unaddressed in the report. A notable 44% emphasized that human oversight must remain a critical component of such access.

The most common agentic AI applications currently in production focus on communication and workflow automation. Fifty-nine percent use AI agents for drafting standardized communications or summaries for stakeholders. Fifty-four percent rely on them for scheduling or routing workflow tasks, such as alert triage and process automation. Other applications include generating standard reports or dashboards without manual intervention (48%), monitoring key performance indicators and triggering alerts (45%), and cleaning or preprocessing routine data sets (45%). Less common uses involve running basic predictive models (34%) or automatically generating insights from data (23%).

Despite this AI adoption, foundational data work remains a significant time sink. Respondents spend nearly six hours per week cleaning and prepping data for AI models or retrieval-augmented generation platforms. Forty-eight percent spend between six and ten hours weekly on these tasks. The tools of choice are still spreadsheets (61%), followed by business intelligence tools (56%) and dedicated data preparation platforms (51%). As the survey authors note, “The continued dominance of spreadsheets reflects a broader reality. AI is layering on top of existing workflows rather than replacing them.”

Another surprising finding is the slow pace of real-time decision-making. Only 20% of organizations report that moving from data analysis to a business decision can happen within a few hours. A mere 5% say they support real-time decision-making.

The biggest barrier to AI adoption? Explaining AI outputs to business decision-makers. Fifty-five percent cite difficulty interpreting or explaining AI results as a major hurdle. Close behind, limited analytical skills among business users (54%) and data that is not sufficiently clean, integrated, or governed (50%) also pose significant challenges. Other barriers include a lack of clarity on ownership or accountability for decisions (49%) and technical limitations of AI tools or infrastructure (45%).

Validation of AI outputs adds another layer of time. Analysts spend nearly four hours per week correcting or validating AI-generated results, with one in six spending six hours or more per week on this task. Combined with the six hours spent on foundational data work, this creates an AI “tax” of almost two days per week. This points to an emerging skill set: validating AI outputs. As the survey authors conclude, it is “a signal that while AI can accelerate work, organizations still need human oversight to ensure outcomes are consistent, explainable, and trusted.”