One in five enterprises can't stop a runaway AI agent's spending in real time
Enterprise AI teams have stopped betting on a single orchestration platform. The median enterprise now runs three at once — not by accident, but because none of them fully trusts a single vendor to run the show, according to VB Pulse data . This is not just to avoid vendor lock-in and retain flexibility (although that’s a big part of it). There’s still a lot of uncertainty, even distrust, in…
One in five enterprises struggle to monitor and control the spending of runaway AI agents in real time. According to VB Pulse data, most enterprises now run three orchestration platforms simultaneously, not out of choice, but due to mistrust in a single vendor's security and permissioning capabilities. Microsoft leads in current usage, while Anthropic is the top choice for future platforms.
However, enterprises face challenges with token usage and visibility into agent spending. When building AI agents, enterprises primarily use Microsoft AI Foundry/Copilot Studio (70%), OpenAI's Agents SDK (68%), and Anthropic's Claude Platform (47%). Additionally, 22% utilize custom in-house orchestration. By the end of 2026, over half anticipate a hybrid control plane, with various providers ranking as top options.
Enterprises prioritize visibility, security, and control, investing in agent monitoring and debugging (31%) and security and permission enforcement (30%). Despite generally high satisfaction with current platforms (4.17/5), ease of implementation (3.91/5) and value for money (3.63/5) are less favorable. Enterprises aim to control agent token use, but one in five cannot stop runaway agent spending in real time, often relying on reactive monitoring or custom gateway plumbing.
Written by urgent.news from VentureBeat's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
