he package also comes with new capabilities to evaluate agent performance before launch, including prompt optimization, evaluations of third-party models and end-to-end assessments of agentic workflows.

In its presentation of the AgentKit, the company successfully built an entire workflow, complete with two working agents, on stage in less than 8 minutes.

With the debut of AgentKit, OpenAI is tapping into an industry trend that executives can’t get enough of: data from IDC, published last week, found that CEOs are broadly bullish on agents, despite the nascent nature of the tech.

Although OpenAI has long held the spotlight in the AI industry, it’s slightly late to the agentic party. Tech firms that have a longstanding footprint in the enterprise market, like Microsoft, ServiceNow, Google and Salesforce, already have their own agentic toolkits on the market. Startups in the space have also started to gain traction, such as n8n reportedly scoring a $2.3 billion valuation in August and Sierra Technologies raking in $350 million in funding at a $10 billion valuation in September.

Still, with OpenAI now boasting a roughly 4 million developer audience, the company’s sheer exposure and name recognition could shoot AgentKit to the top of enterprise competitors. Better late than never.