OpenAI and Hermes: Towards a Revolution in AI Agents with ChatGPT

OpenAI is reportedly developing a new feature for ChatGPT, codenamed "Hermes." This information comes from internal test screenshots shared by Tibor Blaho, a source known for its reliability. Hermes would include a creation tool for agents called "Studio," ready-to-use models, and scheduling functions. These agents could be utilized in ancillary services such as Slack, allowing them to autonomously handle common inquiries or generate weekly reports every Monday at 8 AM.

Hermes promises continuous availability, operating 24/7, with quick setup. The system would also offer the ability to add applications, skills, files, memory, instructions, and other options, making AI agents highly customizable.

Towards a Comprehensive AI Agent Platform

According to other rumors, Hermes would build on existing tools, including custom GPTs, to transform ChatGPT into a platform dedicated to AI agents. The feature would closely integrate with OpenAI's workflow generator, allowing users to deploy their agents within the "Builder" tool. Combined with Codex, this technology could automate many administrative tasks, reducing certain positions to mere supervisory roles or even eliminating them.

No official details or launch timeline have been communicated by OpenAI. Internet users speculate that the company is remaining discreet to ensure it has enough computing power to meet potentially increased demand.

Other Innovations in the Pipeline

Tibor Blaho also mentions other new features in preparation at OpenAI. Among them is "Personalization of Likeness with ImageGen," which uses a reference photo to generate an image. A "Image Contest 2.0" and an "Audio Summary" mode are also mentioned, with the latter offering various formats ranging from public radio-style podcasts to executive briefings and study guides.