As AI technology accelerates, OpenAI is preparing for a radical overhaul of its flagship product, ChatGPT. According to industry insiders, the leading AI research organization is internally testing a revolutionary feature called "Skills," under the project codename "Hazelnut." This move represents a major strategic pivot in AI interaction design and a direct technical response to competitor Anthropic's Claude, which already offers a similarly named capability.
In the current AI application ecosystem, users primarily extend ChatGPT's functionality through customized GPTs (custom GPTs). However, the upcoming Skills feature embodies a completely new design philosophy—shifting from "role-based settings" to "capability-based instructions." This transformation not only reflects the evolutionary trend of AI application development but also addresses the industry's growing demand for large models to move from general-purpose to specialized domains. Skills focuses on teaching AI to master specific abilities, workflows, and domain knowledge, enabling it to serve vertical industry applications with greater precision.
Claude has already demonstrated the advantages of this skill-driven model. For example, its "front-end design plugin" allows the AI to exhibit enhanced perception and professionalism during web application development. Drawing from this success, OpenAI's Skills system is built around four core characteristics that will redefine human-AI interaction. First is composability: skills can be stacked and used together, with the system automatically identifying task requirements and coordinating multiple skills to work synergistically, greatly improving the AI's ability to solve complex problems. Second is high portability: a unified format means developers need to build once, and the skills work across ChatGPT web, desktop, and API, reducing the cost and complexity of multi-platform deployment. Third is extreme efficiency: skills load on demand, activating only when relevant tasks are processed, avoiding unnecessary consumption of context windows and optimizing compute resource usage. Fourth is strong execution: the ability to write and run executable code addresses the unreliability of traditional token generation in logic programming tasks, opening the door for AI to handle more complex computational challenges.
According to industry source Tibor on social platform X, the new feature will introduce a slash command interaction method, significantly boosting operational efficiency and enabling the AI assistant to respond to user commands more quickly. Additionally, OpenAI plans to release a dedicated skill editor, offering developers and advanced users more flexible customization options. Notably, the company will also allow users to convert existing custom GPTs into "skills" with a single click—a smooth migration strategy that lowers user switching costs and accelerates market adoption of the new feature.
From an industry perspective, the launch of Skills reflects the evolution of large AI models from pure conversational interfaces toward intelligent operating systems that integrate multiple capabilities. As AI technology deepens its penetration across various sectors, the demand for professional AI assistants capable of handling complex tasks continues to grow. Skills is designed to meet this market need, enabling ChatGPT to better adapt to enterprise-grade scenarios and become a true productivity tool.