Intelligence Track

Design

24 June 2026

The Brief

The comfortable read is that demand is strong and getting stronger — record spending intent, carriers expanding capacity, experiential budgets outpacing goods. The more uncomfortable read sits one layer down: nearly every signal is also a story about control migrating away from generic matching. The supply growing fastest is the kind that resists price comparison, the visibility that matters most is the kind suppliers can expose directly, the design leverage is moving beneath the interface where an aggregator has never operated, and even marketing is being rebuilt around owned, attributable channels rather than rented reach. Aggregating cheap supply for price-sensitive demand was always the commodity layer — and it is the layer these shifts quietly erode. When demand is this healthy, the question worth sitting with is whether the instinct should be to ride it, or to use the cover it provides to move up into the parts of the journey that won't commoditize.

Hospitality Net · 24 Jun 2026

Argues that appearing in AI assistants is not one problem but three: model memory (training data, least controllable), web search (SEO-adjacent), and connected dynamic data sources (real-time, most controllable). It separates the LLM as reasoning engine from the assistant as product layer that decides whether to answer from memory, search, a connected source or a blend, and flags an emerging paid layer with ChatGPT testing ads and Google's AI Mode advancing direct offers.

Reading as

AI & Design

The layers of AI experience

Proposes a six-layer model of AI experience — interface, context, harness, model, governance, emergence — extending Garrett's Elements of UX and Mill's Product Design framework into probabilistic systems. The argument: because the model itself introduces variance, designers should shape the conditions and leverage points beneath the interface rather than specify every state, and the interface's role shifts from driving the system to overseeing it as onboarding becomes the system learning the user.

If interface work shifts toward oversight and the real leverage moves to context, harness and governance, then design hiring and tooling investment should follow it beneath the surface — toward people who can shape model behaviour, context systems and guardrails — rather than producing more chat-UI patterns.

HeyDesigner·24 Jun 2026