Enterprise AI Adoption Halted by Security and Complexity Concerns

Enterprise Adoption of AI Agents Hinges on Addressing Security and Complexity Concerns

Artificial intelligence (AI) agents are increasingly being integrated into routine business processes, particularly in engineering and IT operations. A recent study by Docker, titled “The State of Agentic AI Report,” reveals that 60% of organizations have deployed AI agents in production environments, with the majority considering agent development a strategic priority.

Initial Deployments and Use Cases

Initial deployments of AI agents focus on internal workflows, with DevOps and Continuous Integration (CI) and Continuous Delivery (CD) optimization being the primary use cases. Security automation and general process automation also rank high, followed by code generation and review. These environments provide structured tasks and measurable outputs, allowing teams to evaluate performance and manage risk.

Security and Complexity Concerns

However, as AI agent deployments scale, security and complexity concerns emerge as significant barriers. 40% of respondents identify security and compliance as the primary obstacle to scaling agentic AI, citing difficulties in verifying that tools meet enterprise security standards. Respondents report issues at infrastructure, operational, and governance levels, including runtime isolation and sandboxing, exposure introduced by coordinating models, APIs, and external systems, and the need for stronger audit mechanisms and consistent policy enforcement.

Operational complexity is another major challenge, with 48% of respondents citing the difficulty of orchestrating multiple components as the primary challenge in building agents. Integrating models, connectors, and runtime environments increases monitoring requirements for security teams, while multi-model architectures raise operational demands.

Model Usage and Deployment

The use of multiple models is widespread, with nearly all surveyed organizations using more than one model within their architectures. 61% combine cloud-hosted and locally hosted models, driven by control, data privacy, and compliance concerns. Hybrid and multi-cloud deployments are also common, with most organizations operating agents in more than one infrastructure environment.

Agent Sharing and Security

Agent sharing practices are fragmented, with commercial marketplaces and source code repositories serving as common distribution channels. Security represents the leading barrier to seamless sharing, with respondents calling for signed and scannable agent packages, centralized registries, and built-in policy enforcement.

Vendor Dependency and Containerization

Concerns about vendor dependency are also prevalent, with 76% of respondents reporting concerns about lock-in related to model hosting platforms, cloud providers, and monitoring layers. Organizations are diversifying models and infrastructure environments to reduce dependency, which increases coordination complexity.

Containers serve as a consistent operational foundation, with a large majority of organizations using containers in agent development or production workflows. Most extend established cloud-native pipelines and orchestration practices to support agent systems.

About Author

Vaibhav

See author's posts