New Research

Agents in Production:
The Builder's Perspective

For many organizations today, AI agents are already in production. But what does it really mean to be "in production" when it comes to agentic systems?

We surveyed 260 AI engineering practitioners and leaders at organizations with 1,000+ employees — the people building and deploying agentic systems in the real world. What they reported is a measurable gap between the speed of deployment and the maturity of operations, with findings that suggest the concept of "production-ready" for agents requires further scrutiny.

260
AI practitioners & leaders surveyed
1,000+
Employee orgs represented
2026
Latest survey data

Download the full report

Inside the report

What we found

The report covers the full landscape of agentic deployment — from how teams define "production-ready" to how they handle it when things go wrong.

 

The state of production deployment

Where organizations actually are in deploying agentic systems — and what "production-ready" means to the people responsible for delivering it.

 

Operational rigor & incident response

How teams are — or aren't — building operational maturity around agentic systems, including incident response readiness and runbook coverage.

 

Where AI visibility breaks down

The specific dimensions — data, behaviors, performance, system outputs — where practitioners lose line of sight into what their agents are actually doing.

 

Accountability when agents fail

How responsibility is structured across engineering, leadership, and cross-functional teams — and where the gaps in ownership emerge.

The findings don't argue for slowing down — they argue for building the operational layer that makes speed sustainable.

— Agents in Production: The Builder's Perspective report

Who this is for

Built for the people building agents

 

AI and ML engineers deploying agentic systems in production environments

 

Engineering leaders accountable for the reliability and performance of AI-powered products

 

Platform and infrastructure teams building the operational layer for agentic workloads

 

Data leaders navigating the intersection of data quality, observability, and AI reliability

Methodology

A rigorous survey of the practitioners and leaders shaping agentic deployment at enterprise scale.

260
AI engineering practitioners & leaders
1,000+
Employee organizations
2026
Survey year
IC → VP
Individual contributors through senior leadership

Get the full report

See what 260 builders reported about agents in production

The complete data, analysis, and implications for teams deploying agentic systems at scale.

Download the Report