Intelligence Track
19 May 2026
Google News (Design AI) · 19 May 2026
Figma is deepening its AI integration across the design platform, targeting faster design production, smarter collaboration features, and improved handoff between design and development. The push reflects Figma's positioning against emerging AI-native design tools.
The article examines how latency in AI-powered products creates a distinct UX challenge — users experience uncertainty and anxiety during wait states that differ from traditional loading patterns. It proposes design strategies for making AI processing feel intentional rather than broken.
Why it matters
Travel search and AI trip planning features inherently involve processing delays; poorly designed wait states erode trust and increase abandonment at a critical conversion moment.
Amplitude built an internal design agent using Claude and Cloudflare that enforces design system consistency across AI-generated UI, addressing the visual incoherence problem that arises when LLMs produce interfaces without brand constraints.
Why it matters
As travel product teams explore AI-generated UI or rapid prototyping, maintaining design system fidelity becomes a technical problem, not just a process one — this case study shows one viable architecture.
The piece argues that AI is shifting the designer's role from craft execution toward experience architecture — defining how AI systems behave across interactions rather than designing individual screens. It frames this as an evolution in scope, not a replacement of design skills.
Why it matters
As OTAs integrate AI into search, recommendations, and support, designers need to evolve their practice toward shaping AI behaviour and guardrails, not just visual interfaces.
Amplitude's piece advises product teams to frame AI queries around specific outcomes rather than open-ended exploration, proposing three outcome-oriented question types for getting useful answers from AI analytics agents.
Why it matters
As travel product and growth teams adopt AI-assisted analytics, developing structured prompting practices will determine how much actionable insight they extract versus noise.