User @j4ck.xyz has demonstrated the use of OpenAI's GPT-OSS model to create a personalized feed on Bluesky. This is a significant development, as it showcases the potential for using large language models to create highly customized and relevant content streams for users. The feed is generated by clustering the user's follows and liked posts, and then using the GPT-OSS model to generate a feed based on these clusters. This approach could be used to create a wide variety of personalized feeds, such as feeds that focus on specific topics, or feeds that are tailored to the user's individual interests. This is another example of the innovative development happening on the AT Protocol, outside of the official Bluesky PBC development.