Okta Uses Internal Dex AI Agent to Test Identity Controls and Target 1 Million Saved Hours

Identity software provider Okta Inc. is using its internal artificial intelligence assistant, Dex, to manage employee support workflows while testing security controls intended for enterprise clients, according to a report from SiliconANGLE during the company's Oktane event.

The company has raised its annual efficiency target to saving 1 million employee hours across the organization after exceeding a previous goal of 250,000 hours, according to Jenna Cline, senior vice president of technology, data and intelligence at Okta.

Speaking on a broadcast with theCUBE, Cline noted that 1 million hours translates to approximately one month of time returned per employee. To hit the target, Okta is combining generative AI systems with established deterministic automated workflows.

To prevent operational fragmentation across internal departments, Okta designed Dex as a front-end orchestrator rather than building an all-encompassing monolithic model or allowing departments to deploy siloed tools. The architecture is intended to prevent a "swivel chair" workflow where employees must navigate separate agent interfaces for different tasks.

Under this system, Dex acts as a single interface that routes requests to scoped sub-agents dedicated to specific domains, including expense processing and IT service desk inquiries.

Okta is also using Dex as a testing ground for its own commercial software. Dex is registered directly within Okta's Universal Directory and incorporates features from the Okta for AI Agents product suite, allowing the company's internal technology team to test and refine features before general availability.