Dynatrace, a leading AI-powered observability platform, has released The Pulse of Agentic AI 2026, a global study examining how observability and reliability determine the successful operationalization of agentic AI.

Based on a survey of 919 senior global leaders responsible for agentic AI implementation, the research reveals that enterprises are not hesitating due to lack of confidence in AI. Instead, they are scaling cautiously because they cannot yet fully govern, validate, or safely scale autonomous systems in production environments.

A Structural Shift: Reliability as the Gating Factor

The research found that roughly half of all agentic AI initiatives remain in proof-of-concept or pilot phases. While adoption is still early, momentum is building quickly, with 26% of organizations already running 11 or more agentic AI projects.

As enterprises move beyond experimentation toward scaled deployment, they are prioritizing platforms that are reliable, trustworthy, and proven in real-world environments. This shift is reflected in growing investment, with 74% of respondents expecting agentic AI budgets to rise again next year.

These findings point to a broader inflection point where reliability, resilience, governance, and real-time insight are becoming defining requirements for enterprise readiness.

Key Findings From the Report

• Nearly half (48%) of leaders expect agentic AI budgets to increase by at least $2 million, signaling continued but measured investment.
• Agentic AI is most commonly deployed in IT operations and DevOps (72%), followed by software engineering (56%) and customer support (51%).
• Improving decision-making through real-time insights is the top priority for 51% of organizations, closely followed by system reliability (50%) and operational efficiency (50%).
• The highest expected ROI areas include ITOps and system monitoring (44%), cybersecurity (27%), and data processing and reporting (25%).
• The leading barriers to production include security, privacy, and compliance concerns (52%), technical challenges in managing agents at scale (51%), and skills or training shortages (44%).

Trust and Human Oversight Remain Central

While enterprises are building toward greater autonomy, human guidance remains a deliberate part of agentic AI strategies. Leaders anticipate a balanced collaboration model, with a 50/50 human–AI split for IT and routine customer support, and a 60/40 split for business applications.

This reflects the continued importance of human judgment in setting goals, defining boundaries, and ensuring accountability as AI agents take on more responsibility.

• 64% of organizations deploy a mix of autonomous and human-supervised agents.
• 69% of agentic AI–powered decisions are still verified by humans.
• 87% are actively building or deploying agents that require human supervision.
• Only 13% rely on fully autonomous agents, while 23% use exclusively human-supervised agents.
• Top validation methods include data quality checks (50%), human review of outputs (47%), and monitoring for drift or anomalies (41%).
• 44% still rely on manual review of agent-to-agent communication flows, highlighting the need for more automated oversight.

“Organizations are not slowing adoption because they question the value of AI, but because scaling autonomous systems safely requires confidence that those systems will behave reliably in real-world conditions.”

— Alois Reitbauer, Chief Technology Strategist, Dynatrace

Observability Enables Trust and Scale

As agentic AI initiatives move beyond pilots, observability is emerging as a critical intelligence layer that provides visibility across the entire lifecycle—from development and implementation to full operationalization.

The study found that observability adoption is strongest during implementation (69%), followed by operationalization (57%) and development (54%), reinforcing its role as a foundational capability for production environments.

• Nearly 70% of organizations use observability to gain real-time insight into agent behavior, system performance, and decision-making.
• 50% deploy agentic AI for both internal and external use cases.
• Half of organizations have agentic AI in production for limited use cases, while 44% report broader adoption across select departments.
• 23% have achieved mature, enterprise-wide integration.

“Observability provides the transparency and confidence organizations need to scale agentic AI responsibly, while maintaining appropriate oversight.”

— Alois Reitbauer, Chief Technology Strategist, Dynatrace

Methodology

The report is based on a global survey of 919 senior leaders and decision makers directly involved in agentic AI development and implementation at enterprises with annual revenues exceeding $100 million.

The research was conducted during November and December 2025 and included respondents from the United States, Europe, Asia Pacific, Latin America, and the Middle East.