Define·Research Access·Agency·Developing·DEF-017
Project Context Assembly
Value hypothesis
Transforms accumulated project knowledge into a queryable intelligence layer, enabling teams to surface relevant precedent, constraints, and insights on demand rather than through manual document search.
Efficiency · Innovation
The designer creates a Project Knowledge Base - an ad hoc but intentionally cultivated collection of project-specific materials (design system rules, heuristics, principles, prior research, user feedback) loaded into a context the designer can query on demand. This is a project-scoped working intelligence layer, not a systematic cross-project repository: it grows through deliberate human curation rather than automated ingestion, and is distinct from the organisation-wide Research Repository covered in IMP-059.
Risks in application
Shallow Solutions
The knowledge base creates confidence in AI-synthesized answers that may be incomplete, outdated, or misrepresent the nuance of source documents.
Black Box Rationale
As AI synthesizes across documents, the specific source reasoning behind an answer becomes opaque; practitioners may accept synthesized guidance without tracing it to its origin.
Expertise that differentiates
Information Architecture
Deciding what belongs in the knowledge base, how to structure it for AI consumption, and what constitutes useful vs. noisy or outdated content.
Research and Insight
Curating research outputs, heuristics, and user feedback to ensure the knowledge base reflects actual evidence rather than assumptions or received wisdom.
AI Fluency that assures
Product Discernment
Loaded into a context the designer can query on demand.
Deciding what belongs in the knowledge base, how to structure it for AI consumption, and what constitutes useful vs. noisy or outdated content.
Curating research outputs, heuristics, and user feedback to ensure the knowledge base reflects actual evidence rather than assumptions or received wisdom.
Creation Diligence
It grows through deliberate human curation rather than automated ingestion.
Related
↓
Enables
Possible Indicators
Query response time
time to retrieve relevant precedent or constraint relative to manual document search baseline
Knowledge utilization rate
proportion of accumulated project knowledge actively surfaced and applied to design decisions
Sources
Cattapan (n.d.). How my design workflow is changing with AI. Medium.
Anthropic (2024). Contextual Retrieval. Anthropic.
Digital Thrive AI (n.d.). Why AI Assistants Don't Share Sources. Digital Thrive AI.
Teeters, R. (2026). Concept to code: Transforming ideas into functional products with Kiro [conference talk]. Designing with AI 2026, Rosenfeld.
Lowson, A. (2026). Rehashing the Double Diamond: collaborating across functions with AI-assisted prototyping [conference talk]. Designing with AI 2026, Rosenfeld.
Oduye, A. (mod.), Crumlish, C., Flowers, E., et al. (2026). From tools to staff: What the next generation of agents means for the future of design [panel]. Designing with AI 2026, Rosenfeld.