Andrej Karpathy has just shared a simple yet effective tip when working with large language models (LLMs): asking the model to format its output as a complete HTML file.
What happened
According to Karpathy, appending the instruction "structure the response as an HTML file" to the end of a prompt allows users to save and open the file directly in a browser to view the results professionally. He also noted success in asking LLMs to present data as presentation slides or other visual formats, rather than just plain text.
Why it matters
For those who frequently use AI for data analysis or content creation, this tip saves time on manual formatting. Turning LLMs into instant UI generators opens up a new approach to consuming AI-generated information, which is particularly useful when quick reports or diagrams are needed.