AI Agent Projects Stall Without Robust Orchestration, Camunda CTO Warns
Table of Contents
despite advances in artificial intelligence, most agent-based AI initiatives remain trapped in the experimental phase due to a critical lack of architectural safeguards for deployment in essential business operations.
The promise of agent-based AI is critically important, but a leading technology executive cautions that realizing it’s full potential hinges on a essential shift in how these systems are built and managed. According to the Chief Technology Officer of Camunda, a holistic platform focused on orchestration is essential to unlock widespread adoption and mitigate risk.
“most agent-based AI projects remain stuck in the pilot stage,” a senior official stated. “Not because the models cannot do it, but because there is currently a lack of an architecture that offers the necessary guardrails to use agents in business-critical processes without risk.”
The Need for Centralized Control
The solution, according to the Camunda CTO, lies in centrally controlling automation and the deployment of AI agents. “A holistic platform is needed to introduce agent-based orchestration company-wide,” they explained. “Agent-based orchestration gives companies control over how much autonomy an agent receives: control where it is indeed needed and flexibility where AI shines.” This approach,combining deterministic processes with the dynamic capabilities of AI agents,is projected to deliver unprecedented productivity gains.
The importance of such platforms is already evident in the market. A recent survey revealed that 77% of organizations are already utilizing automation and orchestration solutions, even among small and medium-sized businesses employing up to 1,000 people. The consistent, end-to-end automation that manny companies seek is simply unattainable without effective orchestration. Furthermore, three-quarters of respondents view this technology as crucial for integrating AI technologies into thier existing workflows.
Real-World Applications of Agentic AI
The potential applications for agentic AI are diverse, particularly in scenarios involving the consolidation of disparate data and processes. LHIND suggests that self-service applications, centrally managed by an agent, represent a prime use case. These applications can streamline tasks across various departments,including human resources and internal services like resource allocation.
Beyond internal operations, AI agents are proving valuable in operational areas. “For one customer, we implemented an agent that bundles different applications for logistics: from the center of gravity model to allocation algorithms for vehicles to trucks, ships or trains,” noted Markus Strittmatter. “The agent knows the applications, knows how to address them, and can carry out tasks for the employees.”
Boosting Customer Service and IT Efficiency
Innovative applications of AI agents were showcased at the Camundacon 2025 conference in Amsterdam. The IT service provider Incentro, for example, developed a customer service agent powered by the Camunda platform. This agent leverages a Large Language Model (LLM) and company data to provide customers with real-time, specific information, reducing inquiry processing times by 50% and increasing customer satisfaction.
In the realm of IT service management (ITSM), professionals are integrating agent-based automation and orchestration with existing systems.Users of ServiceNow’s ITSM platform can now model and execute processes using Business Process Model and Notation (BPMN) and Decision Model and notation (DMN) standards, enhanced by Camunda’s orchestration capabilities. This integration is particularly beneficial for areas like incident management and IT asset onboarding.”By combining BPMN and DMN with AI-supported orchestration, we help our customers reduce manual activities, accelerate IT processes and exploit the full added value of the ServiceNow AI Platform,” the Camunda CTO emphasized.
The future of AI implementation, it appears, is not simply about building smarter agents, but about building the robust, controlled environments they need to thrive.”
