Many organisations still struggle to turn AI pilots into real-world deployments. However, new research suggests enterprise IT teams are making faster progress, with agentic AI increasingly being used in production environments.
Agentic AI adoption accelerates
Agentic AI refers to AI systems that can make decisions and take actions autonomously to achieve specific goals, often with limited human intervention.
According to research from Caylent, a California-based cloud services company and premier AWS partner, all organisations surveyed are actively exploring agentic AI for engineering or operations.
Nearly six in 10 respondents (59.5%) have already moved beyond testing and are using AI agents in production. Within that group, 36% have deployed agents with defined guardrails, while a further 23.5% have rolled them out more broadly across engineering and operations workflows.
Governance remains a top priority
The survey highlights that confidence in autonomous AI depends on strong oversight and control. Almost all respondents said they would allow agents to make production changes autonomously if appropriate safeguards were in place.
In fact, 83% said guardrails are as important as, or more important than, the intelligence and capabilities of the AI model itself.
Current and planned use cases include:
- 67.5% for automated testing
- 60.5% for incident response
- 43% for code generation
The findings suggest organisations are increasingly treating AI agents as operational tools rather than experimental technologies.
How agentic AI creates opportunities for MSPs and partners
While the survey focused on North American organisations with more than 1,000 employees and active cloud workloads, UK and European mid-market businesses are likely to be following a similar path.
For MSPs and cloud partners, growing adoption of agentic AI could create opportunities around deployment management, monitoring, governance, identity management and access control. These areas are becoming increasingly important as organisations look to scale AI safely and securely in production environments.
Caylent’s Enterprise Readiness for Agentic Engineering & Autonomous Cloud Operations Report is based on responses from 200 enterprise business leaders across the US and Canada collected in May 2026.
