ContextOS: Designing for High-Context AI Collaboration

Chat-based LLMs, or AIs, such as ChatGPT, Claude, Gemini, Perplexity, etc., have proliferated in everyday life to provide convenience and improved cognition to their users. Every day more than 120 million users leverage ChatGPT’s capabilities to inform all aspects of their lives from cooking, to data analysis, to personal advice, to data analysis and so much more. However, users today face significant friction when collaborating with LLM chatbots due to fragmented contexts and repetitive tasks, limiting productivity and causing frustration. 

To address these challenges, ContextOS streamlines LLM chatbot interactions by seamlessly integrating fragmented context into workflows, significantly reducing users' effort. Through extensive research involving over 20 participants, we identified critical pain points: fragmented context across various tools, excessive cognitive load from repetitive instructions, lack of precision in contextual relevance, and limited interoperability across platforms. Informed by these insights, our team explored and prototyped a unified, intuitive content management solution, allowing users to easily inject and control context across tasks and LLM chatbot platforms. 

Our solution minimizes workflow disruption, empowers users with granular control, and enables seamless collaboration with multiple AI models. Ultimately, ContextOS dramatically enhances user efficiency and precision, transforming LLM chatbots from mere tools into true collaborative partners.