Abstract
Formulating effective prompts often requires manually curating relevant contextual information, which is a time-consuming and effort-intensive process. Including irrelevant context can degrade the quality of the response, while current automated methods for context selection operate after prompt submission, leaving the user out of the loop and risking incorrect context inclusion. This disclosure describes a method for dynamically suggesting relevant files and other contextual information as a user types a prompt. A lightweight model predicts and displays potential context sources, such as documents, emails, or meeting transcripts, in real-time. Users can then select these suggestions to append them to the prompt before submission. This approach improves prompt formulation by proactively and interactively assisting the user in building accurate and relevant context, leading to higher quality responses.
Creative Commons License
This work is licensed under a Creative Commons Attribution 4.0 License.
Recommended Citation
Barros, Brett and Martin, Ken, "Real-Time Contextual File Suggestion for Prompt Formulation", Technical Disclosure Commons, (September 18, 2025)
https://www.tdcommons.org/dpubs_series/8603