AI agent pilot workflows

last edited

AI agent pilot workflows describe the stage between a test deployment and broad production use. Surveys in 2026 disagree on how many organizations have passed that stage, because definitions of "production" differ. Gartner's figures are the strictest, while vendor surveys report higher shares.

This topic is part of Image Generation Services & Models. Related context includes Technique Adoption Timelines & Standards and Platform Migration Patterns (2024-2026).

Gartner survey figures

Gartner's 2026 CIO survey found that only 17% of organizations have deployed agents. More than 60% expect to deploy them within two years.

The same survey gives 42% planning deployment within 12 months. Another 22% plan it in the following year.

Gartner published its first hype cycle for agentic AI in April 2026. It places agentic AI at the peak of inflated expectations.

Production claims differ

Genpact reported in a survey released on Nov 12 that 79% of enterprises have agents in production. Only 20% use them broadly. The figures are self-reported.

The G2 report on agents covers B2B buyers. It says 66% of them run agents in limited or scaled production. Such surveys count limited use as production, which explains the gap with Gartner's deployment share.

Open model supply

A summer report on open models from the Hugging Face hub shows Qwen leading downloads and derivatives, with 151,448 derivatives. It also names agents as the top user type on the hub.

Some of the largest recent open releases add non-commercial or revenue-share terms, so "open" does not always mean permissive.

Sources