Provide help and guidance that's relevant to the user's current context and AI interaction. This principle ensures that users can get assistance exactly when and where they need it, reducing confusion and enabling successful AI usage.
The Shape of AI framework (Campbell, 2024) identifies contextual help as a key Wayfinder pattern. Generic help documentation fails; assistance that appears at the point of need succeeds.
The finding? Contextual AI help increases task success by 58%—users who receive relevant guidance in context complete their goals far more often than those who must seek help separately.
Interface designers provide AI help effectively. Positioning guidance in context. Adapting content to current task. Making assistance immediately actionable.
The principle: Provide context-aware help. Position guidance at point of need. Enable immediate application.
Contextual AI help has become essential as AI interfaces introduce unfamiliar interaction patterns. Users need guidance that's immediately relevant, not generic documentation.
Campbell's Shape of AI framework (2024) emphasized contextual Wayfinders: "Help should appear at the moment of confusion, not require users to leave their task to find answers."
Microsoft Research (2023) found that in-context help increased AI task success by 58% compared to separate help documentation. Proximity to the task eliminated context-switching costs.
Carroll & Rosson's minimalist instruction research demonstrated that users learn best through immediate application. Contextual help that enables instant trying achieved 42% less abandonment than read-first approaches.
Amershi et al. (2019) noted that AI interfaces benefit especially from contextual guidance because interaction patterns are unfamiliar. Users don't have existing mental models to fall back on.
For Users: Contextual help removes barriers exactly when they appear. Users don't have to remember to seek help or figure out what to search for. Help arrives at the point of friction, enabling immediate progress.
For Designers: Designing contextual help requires anticipating confusion points and delivering relevant guidance. Good contextual help feels like a helpful colleague; poor contextual help is either absent or intrusive.
For Product Managers: Contextual help directly affects completion rates. Users who get stuck and can't find help abandon tasks. Users who receive timely guidance succeed and return.
For Developers: Implementing contextual help requires detecting user state, surfacing relevant content, and delivering assistance without disrupting workflow.
Inline tooltips explain unfamiliar elements. "What's a prompt?" tooltip on prompt input field explains the concept exactly where the user encounters it. Tooltips make unfamiliar elements approachable.
Example prompts demonstrate usage patterns. "Try: 'Summarize this in 3 bullet points'" shows users what good input looks like. Examples are often more helpful than explanations.
Contextual tips appear at decision points. "Not sure what to choose? Most users select..." surfaces when users pause at options. Tips anticipate confusion rather than waiting for explicit help requests.
Error-specific guidance addresses failures. "This prompt was too vague. Try adding more detail about..." turns errors into learning opportunities. Error-time is teaching-time.
Progressive disclosure offers depth for those who want it. "Quick tip" visible immediately with "Learn more" expanding to full explanation. Users get immediate help without being overwhelmed.