[Jung Yeon-tae on AGI] The Door to Personal GPTs Is Narrowing—Who Will Own the AI Agent Era?

The ability to create personalized ChatGPT accounts and publish new GPTs has come to an end, marking a shift in OpenAI's policy. Previously, users could design specialized GPTs for various fields such as history, finance, or medicine using their personal accounts. This change in policy has narrowed the path to developing custom AI agents, with the new focus now on managed Business, Enterprise, and Edu workspaces.

Existing GPTs created under personal accounts remain functional and editable. Users can modify their names, descriptions, conversation starters, instructions, and knowledge files. However, the emphasis has shifted from creating new GPTs to maintaining and enhancing existing ones. The policy change does not indicate any specific business or strategic motives, but it does suggest a broader trend in AI development.

OpenAI's introduction of GPTs in 2023 allowed users without coding skills to build AI systems tailored to specific purposes. This new approach opened the door for individuals with knowledge to create AI systems, from medical GPTs for doctors to educational GPTs for teachers and historical GPTs for historians. The success of GPTs hinged not only on the sophistication of prompts but also on the accuracy and structure of the knowledge, as well as the reliability of sources and the verification process.

The transition from personal GPT creation to agent-based workflows implies a shift from creating individual chatbots to designing and managing organizational tools. GPTs will likely become integral components of larger systems, working alongside agents that perform tasks, use connected apps, and run according to schedules or API triggers.

While the ability to create new personalized GPTs through personal accounts is no longer available, individuals can still develop AI services using APIs, open-source models, and external agent frameworks.