Enterprises are increasingly recognizing artificial intelligence (AI) as a strategic differentiator, with agentic AI – systems capable of autonomously planning and executing multi-step actions – emerging as the next frontier in IT operations. This shift promises unprecedented efficiencies and problem-solving capabilities. However, a recent study indicates a significant disconnect between this ambition and current organizational readiness, alongside pervasive concerns regarding trust, governance, and foundational technical capabilities.
According to the Riverbed Global Survey 2026, The State of Autonomous IT Operations, which surveyed 1,200 business decision-makers, IT leaders, and technical specialists across seven countries, 100% of organizations consider agentic AI strategically important. Yet, most are not prepared to implement it effectively. This article will explore the prevailing ambition, the critical barriers to adoption, and the essential steps for establishing robust AI observability and governance to realize truly autonomous IT operations.
The Strategic Imperative and Its Unmet Foundations
The drive toward agentic AI is clear. Businesses globally report strong returns on their AI investments, with 91% stating that their AI projects in 2026 met or exceeded original goals. This success fuels a unanimous expectation: 90% of companies intend to leverage agentic AI for autonomous IT operations within the next two years, with 21% aiming for highly autonomous, self-managing systems.
However, this ambition confronts a significant readiness gap. Currently, only 19% of IT operations are automated, and just 41% of organizations possess an IT environment fully prepared for autonomous operations. A key impediment is a pervasive lack of trust: 77% of organizations express hesitation about allowing AI to make operational decisions without human approval, a sentiment that has increased since 2025. This indicates a preference for human-supervised automation (69%) over fully self-managed systems (21%).
Furthermore, a notable divergence exists between leadership and technical specialists. While business and IT leaders are more optimistic about agentic AI’s potential, technical specialists—those responsible for deployment and operation—are more acutely aware of the practical hurdles. This internal misalignment can delay critical investments and undermine implementation efforts.
What this means: The strategic intent for agentic AI is firm, but its operationalization is hampered by a significant gap in foundational readiness and a prevailing need for human oversight. Bridging this gap requires addressing the underlying technical and organizational challenges to build trust and competence.
Critical Barriers to Realizing Autonomous IT
Operationalizing agentic AI for autonomous IT is obstructed by several interconnected issues, primarily related to data quality, fragmented observability, and tool sprawl, all compounded by broader security and trust concerns.
The Riverbed survey identifies key challenges:
- Security and Compliance Concerns: Cited by 55% of respondents as a major blocking issue.
- Risk of Operational Disruption: A concern for 45% of organizations.
- Lack of Trust in AI Decisions: Identified by 39% as a significant barrier.
- Skills Gaps: Affecting 37% of enterprises.
- Disconnected Tools and/or Vendors: A challenge for 35%.
- Organizational Resistance to Change: Noted by 34%.
- Poor Data Quality: Cited by 30%.
Three foundational issues stand out as critical technical and operational barriers:
- Trusted, AI-Ready Data: Only 34% of businesses have full confidence in the suitability of their data for agentic AI decision-making. This confidence has declined since 2025, suggesting that as AI initiatives advance, underlying data quality issues become more apparent. Less than a quarter of respondents trust their data’s quality (21%), completeness (33%), or availability for integration across domains (25%). Without high-fidelity, real-time data, AI agents cannot make accurate, reliable decisions or effectively resolve issues.
- Unified Observability: While 96% of organizations agree that multi-domain visibility (across networks, applications, endpoints, and cloud) is essential for AI-driven IT operations, only 17% have achieved fully unified visibility across all domains. This fragmented visibility prevents AI agents from having the complete, contextual information needed to diagnose root causes and take effective action autonomously. Technical specialists, in particular, report lower levels of unified visibility than leaders, highlighting a critical operational disconnect.
- Consolidated Technology Stack: Tool sprawl remains a significant challenge. Although 91% agree that tool consolidation would standardize processes and reduce operational friction, 87% still need to integrate the disparate systems they rely on to enable AI success. A fragmented tech stack increases complexity, adds cost, and impedes AI agents’ ability to access necessary data and workflows efficiently.
What to do:
- Data Strategy: Implement a comprehensive data governance framework focusing on data accuracy, completeness, and real-time availability. Establish data ownership, implement automated data quality checks, and define data freshness SLAs (e.g., critical operational data updated within 60 seconds).
- Unified Observability Platform: Invest in a unified observability platform that aggregates full-fidelity data across all IT domains (network, application, cloud, user experience, infrastructure). Standardize data collection and telemetry using industry standards like OpenTelemetry to ensure contextualized insights for AI agents. Aim for a unified visibility score of at least 80% across critical business services.
- Technology Stack Rationalization: Conduct an audit of the current IT operations tool stack to identify redundancies and opportunities for consolidation. Prioritize unified platforms from fewer vendors to reduce integration complexity and operational overhead. Target a 25% reduction in redundant tools within 12-18 months.
Governance, Trust, and the Path to Accountable Autonomy
As AI and agentic capabilities become embedded within IT operations, organizations must acknowledge the new imperative for robust AI observability and governance to mitigate risks, build trust, and ensure accountability.
The vast majority (92%) recognize that AI observability and governance will become a critical new IT domain. Key requirements for expanding AI use include:
- Monitoring AI application and agent performance (81%).
- Measuring AI usage across teams and individuals (78%).
- Understanding AI costs and resource consumption (77%).
- Resolving AI issues before they impact employees (76%).
- Tracking the adoption of sanctioned AI tools (75%).
- Detecting shadow AI and unsanctioned tool usage (75%).
Despite these recognized needs, current monitoring capabilities are insufficient. Half (51%) lack adequate oversight of AI agent performance and reliability, 49% are unaware of how employees are adopting and using AI, and 44% do not sufficiently monitor AI costs. This gap is further highlighted by a confidence disparity: 48% of leaders are confident in their organization’s AI governance, compared to only 36% of technical specialists. This suggests that those directly responsible for managing AI operations have a more realistic, and often less confident, view of current control mechanisms.
Immediate Priorities (First 90 Days):
- Establish AI Governance Council: Form an interdisciplinary AI governance council involving IT, legal, compliance, and business leaders. Define clear policies for AI agent deployment, ethical use, data privacy (e.g., data handling policies adhering to GDPR), and decision-making thresholds (e.g., automated actions requiring human approval for changes impacting over 10% of users).
- Implement AI Observability Tools: Deploy or extend existing observability platforms to monitor AI agent performance, reliability, resource consumption, and decision outputs. Establish AI-specific SLAs (e.g., an AI agent must maintain a P95 success rate of 99% for automated incident resolution, with a response time under 100ms) and RAG statuses for automated processes.
- Initiate Shadow AI Audit and Policy: Conduct an audit to identify unsanctioned AI usage and tools. Develop and communicate a clear policy for vetting and approving new AI technologies, including an escalation path for non-compliance (e.g., immediate disablement of non-compliant tools, mandatory training for users).
What to Avoid:
- Deploying AI without guardrails: Do not implement autonomous AI agents without defining clear operational boundaries, human override protocols, and automated rollback capabilities for critical functions (e.g., system configuration changes, financial transactions).
- Ignoring technical specialists’ concerns: Overlooking the skepticism and practical insights of technical teams regarding AI readiness, security, and operational risks will lead to failed deployments and erode organizational trust in AI initiatives.
- Optimizing for a single metric: Avoid prioritizing AI efficiency or containment (e.g., fully automated service desk interactions) at the expense of broader outcomes such as system reliability, security posture, or end-user experience (e.g., do not target a 90% automation rate if it increases critical incident recurrence by 15%).
Summary
The journey toward autonomous IT operations, powered by agentic AI, presents a transformative opportunity for enterprises. However, this path is not without its complexities. The Riverbed Global Survey 2026 underscores that while the ambition for AI-driven autonomy is universal, readiness lags significantly. Success hinges on addressing foundational challenges related to data quality, achieving unified observability across fragmented IT environments, and strategically consolidating the technology stack.
Crucially, building trust in autonomous systems requires robust AI observability and governance. By establishing clear policies, continuously monitoring AI agent performance and usage, and fostering alignment between business and technical leadership, organizations can navigate the risks and unlock the full potential of agentic AI. This strategic approach ensures that autonomous IT operations contribute measurably to business outcomes while maintaining control and accountability.
Source: Riverbed Global Survey 2026, The State of Autonomous IT Operations, published by Riverbed.
