Anthropic Claude leads enterprise orchestration, but 71% of 'agents' are chatbots

VentureBeat's Pulse Research surveyed 101 enterprises (100+ employees) on agent orchestration. The central finding: a major gap between ambition and reality. Anthropic's Claude is the primary platform for 40% of respondents, far ahead of Microsoft (18%) and OpenAI (13%). Choice is driven by 'model gravity' (21%) — native alignment with a leading base model — and success is judged by reliable multi-step execution (task completion reliability 32%, multi-step workflow management 28%). Yet when asked to assess their own deployments, 71% say a quarter or fewer of their so-called 'agents' are true multi-step orchestrated workflows; most are single-prompt chatbot wrappers. Only 10% have crossed the halfway mark. This means organizations are building orchestration layers well ahead of the actual orchestrated portfolio they need to run.

Architecture expectations reflect the desire to avoid lock-in. By end of 2026, 51% expect a hybrid control plane (provider-native plus external orchestration), while only 6% plan to hand full control to a provider-managed service. Vendor lock-in is the top risk (35%). Investment follows: agent workflow tooling leads spend at 34%, with security and permissions enforcement at 25%. Fiscal control remains a blind spot — 27% have no real-time way to stop a runaway agent before the bill arrives. The research underscores that enterprises are consolidating on major model platforms but must address the gap between chatbot wrappers and true multi-step agents to realize value from their orchestration investments.