OpenAI's Codex has unveiled a new feature called Appshots, designed to enhance the way users interact with AI by allowing them to directly share what is currently on their screen. This development marks a significant advancement in making AI tools more integrated and contextually aware of user workflows.
Related startups
The Appshots feature allows Codex to process visual information from various applications. For instance, a user can share an email containing event details, and Codex can then interpret this information to add the events to a calendar. Similarly, a user can share a photograph, and Codex can be prompted to modify it, such as transforming it into an anime-style image. The system also demonstrated its ability to ingest design mockups from Figma and generate a press release based on the visual content.
The full discussion can be found on OpenAI Youtube's YouTube channel.
This functionality extends to more complex documents as well. In one demonstration, Codex processed a fundraiser brainstorm document, offering to clean it up and convert it into a Google Doc. This highlights the versatility of Appshots in understanding and acting upon structured and unstructured visual data from a user's digital environment.
