All posts
ai-agentsproductionsecurityobservability

Agent Fleets Doubled in Four Months. 90% Are Running Unmonitored.

750 tech leaders were surveyed on what's actually in production. The fleets grew. The instrumentation didn't. Half of every agent estate is dark.

NeuroX AI · August 2, 2026

Gravitee surveyed 750 senior technology leaders across the UK and US in April 2026, and one number does most of the work: 90% of organizations have unmonitored agents running in production. Not unmonitored pilots. Production.

The context explains it. Agent estates doubled in the four months since December 2025, and 38% of respondents now run more than 100 agents. Mean monitoring coverage sat at roughly 52% — so half the fleet is dark, and the fleet is compounding faster than anything wrapped around it. 48% of production agents are running unsecured outright.

Ownership follows the same curve. Only 7.2% of organizations have a named individual formally accountable for agent behavior; 85% have no formal accountability at all. Meanwhile 81% report pressure to deploy fast.

The bill is already arriving. Cloud Security Alliance and Token Security research found that 65% of organizations had at least one security incident in the past year caused by an AI agent on their network — 61% of those involving sensitive data exposure.

The gap here isn't capability. The agents work. The gap is that teams keep shipping an agent like a feature when it behaves like a deployment. A deployment has an owner, a log, an alert, and a kill switch. If you can't say who owns an agent and what it did at 3am, you don't have 100 agents — you have 100 untracked processes holding credentials.

See how we ship agents you can actually operate →

Contact

Working on something similar?

Tell us about it — we reply within one business day.

Or skip the form — book a Calendly slot directly

We reply within one business day · NDA on request

admin@neuroxai.com · +91 70149 99768

Remote-first team across India · US · EU · HQ in Udaipur, India