Automatically capture relevant context to enhance user inputs without requiring manual specification. This principle ensures that AI has the information it needs to provide relevant outputs while minimizing the burden on users to provide that context.
The Shape of AI framework (Campbell, 2024) identifies context capture as a key Input pattern. Users often don't know what context AI needs; intelligent systems capture it automatically.
The finding? Automatic context capture improves AI output relevance by 56%—when AI understands the user's situation, it provides far more useful responses.
Interface designers enable context capture effectively. Detecting relevant context. Capturing it transparently. Enhancing inputs intelligently.
The principle: Capture context. Enhance inputs. Respect transparency.
Context capture has become critical as AI systems handle complex, situation-dependent queries. Users often can't articulate all relevant context, yet context dramatically affects output quality.
Campbell's Shape of AI framework (2024) emphasized automatic context: "The best AI interfaces capture context invisibly, reducing user burden while improving output relevance."
Stanford HAI research (2023) found that automatic context capture improved output relevance by 56% compared to context-free interactions. Users received more useful responses without extra effort.
Horvitz (1999) established principles for mixed-initiative systems that proactively gather context. His research showed that intelligent context use reduced clarification rounds by 38%.
Amershi et al. (2019) noted that context capture must balance utility with privacy. Users should understand what context is being used and have control over it.
For Users: Context capture means less work for better results. Users don't have to explain their entire situation—AI understands from context. Less effort plus better output equals higher satisfaction.
For Designers: Designing context capture requires identifying relevant signals and presenting them transparently. Good context capture feels helpful; poor context capture feels invasive or creepy.
For Product Managers: Context capture directly affects perceived AI intelligence. AI that "gets" the user's situation feels smart. AI that ignores obvious context feels dumb.
For Developers: Implementing context capture requires identifying relevant context sources, accessing them appropriately, and incorporating them into AI processing.
Current document/page provides immediate context. "AI, summarize this" works because AI knows "this" is the open document. Immediate context is the most common and valuable capture.
Selection defines focus. When users select text before invoking AI, the selection is the focus. Selection context tells AI what specifically the user cares about within larger content.
Conversation history maintains continuity. Previous exchanges in a conversation provide context for follow-up questions. History context enables natural "it," "that," and "more" references.
Application state provides situational awareness. Which screen is open, what filters are active, what the user just did—application state provides rich context for relevant responses.
User preferences enable personalization. Past behavior, stated preferences, and user settings allow AI to tailor responses to individual users. Preference context improves relevance over time.