Finding
The first job: discovery.
The job
Traveller
Job: Find somewhere that fits my occasion
I'm planning a trip, a meal, an experience. I need to find options that match what I'm looking for—not just availability, but fit.
Venue
Job: Be found by the right guests
I need the guests who'll appreciate what we offer. Not just any booking—the right booking.
What's broken today
Traditional discovery forces travellers to translate intent into keywords. "Anniversary trip, quiet, somewhere we can walk to dinner" becomes searches for "hotels Edinburgh" followed by hours of scrolling photos and reading reviews.
For travellers:
- Search results optimised for keywords, not intent
- No way to express what an occasion feels like
- Discovery rewards marketing spend, not fit
For venues:
- Invisible to AI agents that can't read websites
- Competing on SEO and ad spend rather than quality
- The guests who'd love you can't find you
What changes with agents
We're entering post-website hospitality. The traveller doesn't visit booking sites—they tell an agent what they need and the agent finds options. This inverts discovery entirely.
From searching to specifying: Travellers describe their occasion in natural language. "We're celebrating our 10th anniversary, want somewhere romantic and quiet, dog-friendly, walkable to restaurants." The agent understands intent, not keywords.
From venues to ecosystems: Agents don't visit individual websites. They query a structured catalog where every venue has an AI-readable identity. Venues that aren't in the catalog don't exist to agents.
From marketing to matching: Discovery stops rewarding ad spend and starts rewarding fit. The venue that's genuinely perfect for an anniversary trip appears—not the one that bid highest on "romantic hotel."
How the specs answer this
| Spec | What it does |
|---|---|
Venue.identity | Verifiable existence: name, location, domain |
Venue.vibe | Subjective character structured for matching |
Curator.coverage | What the curator knows about—venues, geography |
Curator.actions.recommend | How agents ask for recommendations |