New surveys among more than 100 large organizations show that Agentic Orchestration is rapidly centralizing on major model provider platforms. Among them, Anthropic’s "Claude" model is leading the competition by a wide margin. However, business ambitions are moving far ahead of technical realities; most of what is currently deployed in organizations as "AI Agents" is actually nothing more than simple chatbot wrappers.
The key findings of the VentureBeat report indicate a deep gap between the desire for automated process management and the existing reality. Currently, 40% of organizations use Anthropic’s platform to manage their agents, which is more than double its nearest competitor, Microsoft (18%), and Google. However, what is striking is that 71% of executives admitted that less than a quarter of the agents deployed in their organizations have the capability to execute true multi-step workflows.
One of the primary concerns for IT managers is the fear of "vendor lock-in." For this reason, more than half of organizations expect to develop their control structures in a hybrid format by the end of 2026 to both leverage the power of ready-made models and maintain data sovereignty and system control. Furthermore, cost management challenges persist; 27% of companies have no real-time monitoring tools to stop an intelligent agent in the event of excessive token burn and only discover the heavy costs after receiving the bill.