Intelligence Track
20 May 2026
The Brief
Today's signals converge on one theme: the booking funnel is being attacked from both ends of the stack.
At the top, Google has formally named hotels as its next agentic vertical and Airbnb has crossed into ground transport — two moves that compress the space where OTAs justify their margin. At the bottom, Air India's fleet contraction and the strained economics of its cabin refresh are concentrating Indian domestic supply into IndiGo's hands and opening international corridors to foreign carriers, which changes the inventory mix every Indian OTA has to merchandise. The Figma agent launch and the friction-in-travel argument are not side stories — they are the design-side response to the same compression: design systems become the lever that decides which teams can move at agent speed, and friction becomes a design variable to defend rather than an inefficiency to remove. The read for the team is that the next planning cycle has to make two decisions simultaneously — where Cleartrip stands in the agentic interface layer above the OTA, and how it absorbs an India aviation mix that is structurally less diverse than the one its product was built for.
Google News (Design AI) · 20 May 2026
Figma has released an in-product AI agent that operates on a file with awareness of components, variables, and design intent, going beyond suggestion-style assistance into autonomous task execution inside the canvas. The launch signals Figma's shift from collaborative tool to agentic design platform.
Day one of Google I/O 2026 brought a new Gemini 3.5 model family, an updated Anti-Gravity 2.0 agent framework, and a set of builder-facing tools positioned for developers shipping AI features. The recap separates announcements that are genuinely usable now from ones that are demo-stage.
Why it matters
Product and design teams evaluating AI vendor commitments need to factor in Google's renewed builder push, because Gemini's pricing, latency, and agent affordances now compete head-to-head with the OpenAI and Anthropic stacks that most travel teams default to.
The argument is that two decades of frictionless-first product orthodoxy has eliminated checkpoints that did real work — confirming intent, surfacing tradeoffs, reducing regret — and the AI era will expose which of those removed steps were structural rather than ornamental. The piece reframes friction as a design variable to tune, not a defect to remove.
Why it matters
Booking-funnel teams optimising for fewer taps may now have to defend which steps still exist on purpose, especially as agentic flows compress decisions the user used to make consciously.