OpenAI pushes ChatGPT Work beyond coding

OpenAI is betting that ChatGPT Work can move AI agents from software engineering into everyday office workflows. The product, released last month on the company’s lowest subscription tier for $20 a month, adapts ideas from Codex so non-engineers can ask ChatGPT to complete multistep tasks across tools such as email, Slack, calendars, cloud drives and SaaS platforms.

The opportunity is clear, but adoption remains uneven. An OpenAI-backed study found that in June, 98% of OpenAI employees were using Codex, compared with 17% of organizational subscribers and less than 1% of individual subscribers. OpenAI is trying to close that gap with a more approachable interface, visible controls for projects and plug-ins, and benchmark testing based on GDPval, which draws from 44 occupations and hundreds of knowledge work tests.

The product also exposes the trade-offs behind agentic software. Useful agents need broad permissions and personal context, raising trust and privacy concerns, while setup can be confusing and some tasks still require high effort settings to perform well. OpenAI argues that better models and simpler harnesses will reduce complexity over time, even as rivals, open source developers and analysts question whether model-specific agent apps can sustain an advantage.