Enterprises Utilize Multiple AI Agent Orchestration Platforms
Companies employ an average of 3.1 AI orchestration platforms simultaneously, with a significant portion lacking real-time cost control for agents.
Enterprises are adopting a portfolio approach to AI agent orchestration, utilizing an average of 3.1 platforms concurrently, according to new VentureBeat Pulse Research.
The survey of 107 enterprises reveals that flexibility across various AI models is the primary driver for platform selection (29%), outweighing affinity to any single model. Security and permissioning (17%) are also key considerations. Microsoft platforms lead in current usage, while Anthropic shows strong consideration for future deployments.
Looking ahead, a hybrid control plane combining provider-native and independent technologies is anticipated. Companies express greater concern over provider-specific security and permissioning limits (37%) than vendor lock-in. Investments are prioritizing agent monitoring and debugging (31%) and security enforcement (30%).
Many organizations admit a majority of their 'agents' are essentially chatbots. Only 16% report over half their agents as genuinely orchestrated. A critical gap remains in fiscal control, with 21% lacking real-time methods to halt runaway agent executions before incurring significant costs.