Source: 10’000 Hours/DigitalVision via Getty images.
A study conducted by 451 Research from S&P Global Energy was designed to assess how organizations are adopting agentic AI and to identify emerging opportunities, risks, and best practices shaping its use in business and IT operations. Conducted online in Q4 2025, the survey gathered responses from IT decision-makers, practitioners, and influencers across industries such as software and IT services, manufacturing, healthcare, finance, retail, and government/education. The respondent pool was diverse in company size, revenue and job roles, providing a comprehensive snapshot of current trends and future intentions in AI adoption.
The findings show that agentic AI adoption is accelerating, with many organizations moving from experimentation to production, particularly in software and IT services. However, deployment remains measured, as organizations prioritize governance, auditability and explainability. Senior decision-makers are driving adoption, favoring embedded vendor platforms while balancing innovation with risk mitigation, human oversight and controlled scaling.
The Take
The survey results reveal an organizational landscape moving at a surprising speed toward a “multi-agent” operational reality. The fact that the average organization already manages 19 agents and plans to double that number within a year indicates that agentic AI has moved past the “hype” phase into rapid industrialization. The heavy reliance on incumbent vendor platforms (65%) suggests that the first winners in this space will be the established giants who can seamlessly add “agentic layers” to their existing products. However, the high percentage of organizations building their own tools (51%) shows a persistent need for customization that off-the-shelf solutions may not yet meet.
The most critical data point is that 36% of organizations insist on human-in-the-loop oversight for mission-critical processes.
This highlights a “trust gap” that remains the primary bottleneck for true autonomy. While the technical ability to deploy agents is scaling exponentially, the organizational ability to trust those agents without constant surveillance is still developing. Organizations that can master “automated guardrails” – moving from manual human intervention to automated policy enforcement — will likely be the ones to see the most significant competitive advantage in the coming years. For now, the “supervised agent” is the organizational standard.
Summary of findings
Organizations aggressively adopt agentic AI within core operations. The survey results indicate that a substantial majority of organizations are already actively engaging with agentic AI technologies. Currently, 43% of surveyed organizations have these tools in active use, while an additional 30% are navigating the proof-of-concept stage. This combined figure suggests that nearly three-quarters of the organizational market is moving beyond theoretical interest into practical application. Furthermore, 12% of respondents plan to initiate use within the next 12 months, leaving only a small fraction of 10% with no current plans or awareness of agentic AI initiatives. This rapid uptake underscores a fundamental shift in IT strategy, where autonomous agents are becoming a standard component of the digital transformation tool kit rather than a peripheral experiment.
Software and IT services outpace other industries in agentic AI adoption. Software and IT services organizations lead agentic AI adoption, with 50% currently using the technology, 29% in the proof-of-concept stage and 11% planning to adopt within 12 months. By comparison, Healthcare adoption stands at just 28%, while government/education is at 29%. Manufacturing and business services show moderate uptake, with 32% and 35% using agentic AI for operational efficiency and automation, respectively. This disparity reflects sector-specific priorities — innovation and rapid iteration in software and IT, versus regulatory and risk concerns elsewhere. Organizations in slower-adopting sectors can learn from the strategies and results achieved by software and IT services leaders.
Decision-makers lead the strategic design of agentic AI platforms. Strategic oversight of agentic AI is concentrated among high-level stakeholders, with 51% of respondents identifying as decision-makers who create strategy and actively design or manage these platforms. This high level of executive and managerial involvement suggests that agentic AI is viewed as a foundational strategic asset rather than a localized technical tool. Meanwhile, 18% of participants serve as practitioners who develop and support applications using these technologies and 16% serve as influencers who provide critical input on projects. The prominence of decision-makers in the survey pool highlights that agentic AI initiatives are being driven from the top down, ensuring that deployment remains aligned with broader corporate objectives and long-term digital infrastructure planning.
Mature DevOps frameworks support the integration of autonomous agents. There is a clear correlation between DevOps maturity and the readiness to implement agentic AI, with 49% of respondents reporting full DevOps adoption across 100% of their IT organizations. Another 45% indicate some adoption at the team level, meaning nearly 94% of the organizations surveyed have at least a partial DevOps foundation in place. This environment of rapid software iteration and collaborative operations provides the necessary infrastructure for AI agents to thrive. Only a negligible 2% of organizations report no DevOps adoption. The prevalence of established DevOps practices suggests that most organizations already possess the cultural and technical prerequisites to managing the continuous deployment and monitoring cycles that autonomous agents require.
Rapid software delivery cycles demand increased agentic automation. The frequency of software deployment among respondents further validates the need for automated agentic support. Approximately 33% of organizations deploy software applications to production weekly, while 26% deploy daily. An additional 5% of the most agile organizations are deploying hourly. In total, over 64% of organizations are pushing code to production at least once a week. This relentless delivery pace creates significant pressure on manual operations and traditional automation, driving demand for AI agents capable of making autonomous decisions and managing complex tasks without constant human intervention. As deployment windows continue to shrink, the role of agentic AI in maintaining stability and performance becomes increasingly critical.
Organizations achieve full-scale adoption across many business divisions. Current deployment patterns show that agentic AI is quickly moving past isolated silos. Among organizations currently using or piloting these tools, 35% have achieved full-scale adoption, with utilization organization-wide. Another 31% report broad-scale adoption across many areas or divisions, even if it has not yet become a universal standard. This indicates that for 65% of active users, agentic AI is already a significant operational component rather than a niche experiment. Only 11% of respondents categorize their use as very limited or experimental. This data point is particularly telling, as it suggests that once an organization moves past the proof-of-concept stage, the technology tends to proliferate rapidly across various functional departments.
Future expansion plans target specific, segmented functional projects. When looking toward the next 12 months, organizations currently in the planning phase anticipate a more measured rollout compared to early adopters. Nearly half of these planners, at 47%, expect “some adoption” characterized by select teams or departments using agents for specific functions or projects. About 28% anticipate broad-scale adoption across many divisions, while 17% are aiming for full-scale organization-wide implementation. This forward-looking outlook suggests that the next wave of adopters may be taking a more incremental approach, focusing on high-value use cases and departmental successes before committing to an all-encompassing rollout. This phased strategy likely reflects a desire to validate ROI and refine risk mitigation protocols within specific business contexts.
Substantial agent footprints are emerging in the current organizational landscape. The density of AI agents within organizations is already significant, with the survey revealing a mean of 19.2 agents currently deployed per organization. The median stands at 18 agents, indicating a relatively consistent level of deployment across the respondent base. Within this distribution, 19% of organizations have 6-10 active agents, while 17% have 16-20. Interestingly, a notable 7% of organizations have already surpassed 40 active agents. This volume of deployment suggests that organizations are not just experimenting with a single “assistant” but are instead building out fleets of specialized agents to handle a diverse array of tasks across business and IT operations.
Projected inventories indicate a scaling of agentic capabilities. The momentum of agentic AI is set to accelerate, with organizations expecting to deploy a mean of 19.4 new agents over the next 12 months. This essentially represents a doubling of the current average agent footprint. Among those planning expansion, 15% expect to add 6-10 new agents, while 7% plan to add more than 40 new agents in the coming year. This aggressive growth trajectory indicates a high degree of confidence in the value delivered by early deployments. As organizations become more comfortable with the management and orchestration of these autonomous entities, they are clearly moving toward a “multi-agent” future where dozens of autonomous systems work in concert.
Strategic alignment, integration and implementation speed dictate the priority of use cases. When prioritizing use cases for current agentic AI deployment, organizations focus heavily on strategic fit and technical efficiency. Strategic alignment leads the list of influencing factors at 51%, followed closely by integration improvement (31%) and the ease of implementation, or “quick wins,” at 29%. Potential efficiency gains (28%) are also top-tier considerations. For those in the planning stages, integration remains the dominant priority, cited by 40% of respondents. This data suggests that the immediate value of agentic AI is found in its ability to bridge disparate systems and automate workflows that were previously difficult to connect. Organizations are favoring projects that offer immediate operational relief and can be woven into existing tech stacks.
Human oversight remains central to organizational risk mitigation strategies. Despite the autonomous nature of AI agents, organizations are emphasizing human governance to mitigate risks such as misinformation and errors. Currently, 38% of organizations monitor autonomous actions and validate decisions against guardrails in real time. Crucially, 36% have implemented human-in-the-loop oversight specifically for mission-critical processes and 28% require human intervention for all processes. Other popular measures include logging decisions for audit trails (34%) and limiting AI autonomy to low-risk tasks (32%). This cautious approach to autonomy demonstrates that, while organizations value the speed of AI, they are unwilling to compromise on reliability and accountability, preferring a “supervised autonomy” model in which humans remain the ultimate authority.
Vendor-embedded solutions anchor the majority of current deployments. The market is currently leaning toward established providers for agentic AI delivery. Approximately 65% of organizations use agentic AI tools embedded in their incumbent IT vendor platforms. This approach likely offers a lower barrier to entry and better compatibility with existing workflows. However, the market is far from a monoculture; 55% of respondents are also using independent, specialized agentic AI designer and orchestrator platforms, and 51% are developing their own tools in-house. These overlapping percentages indicate that many organizations are pursuing a hybrid strategy, using vendor tools for standard tasks, building custom solutions for proprietary needs or leveraging specialized platforms for more complex orchestration.
Organizations prefer consolidated vendor platforms for long-term growth. When asked about their most preferred approach for the future, the trend toward vendor consolidation becomes even clearer. Nearly half (46%) of respondents identify agentic AI embedded within incumbent IT vendor platforms as their preferred deployment method. Independent specialized platforms are favored by 31%, while 21% prefer developing their own tools. This preference for embedded solutions suggests that, as the market matures, IT leaders value the convenience of integrated ecosystems and the support of established vendors over the flexibility of bespoke development. For vendors, this signifies a massive opportunity to retain and expand their footprint by successfully integrating agentic capabilities into their existing software suites.
HR Transformation: from Compliance to Culture Architect
Want insights on workforce productivity and collaboration trends delivered to your inbox? Join the 451 Alliance.
