Realty Connect Global Summit 2026. 8,500 attendees. Three days. You walk in. The app doesn't open with a menu — it opens with a question. You answer in plain words. It guides you to the right person, the right session, the right room. No tab bar. No feature grid. No hunting.
Sara opens the app at 10:24 AM. She's been awake for two hours and hasn't talked to a single person yet. The app doesn't ask her to scroll a list of 8,500 names. It asks her one question, and her answer is the first move of the day.
What would you like to do?
The home screen is one greeting and three suggested prompts. No tabs. No notifications panel. One question, three answers.
"Who should I meet now?" — a question she'd say out loud to a colleague. The app receives it like one.
Not 47 names ranked by a fit score. Three people. Each card answers the only question that matters: why this person, right now?
The connection is made. The app immediately offers the next step before Sara has to ask: where to meet, how close, how long.
30 feet. One minute. The map is a hint, not a feature. The conversation has carried Sara from "good morning" to "shaking hands" in eighteen seconds.
Same app. Different question. Sessions are not the primary surface — they're a guided answer when Sara asks. No long lists, no filters, no track-by-track tabs. The app picks one session, says why, and offers the next step.
What's on your mind?
The same conversational entry, two hours later. Sara doesn't open an "Agenda" tab. She asks.
Not a list. One session — because Sara asked what's next, not what's available. The app shows why.
The phrase is the action. No "register," no "RSVP," no toast that says "added to calendar."
Anna is also at the 6 PM mixer. Sara didn't ask about Anna twice — but the app remembers Sara liked Anna, and offers the next moment to meet her in person.
Sara's profile is what the app uses to suggest who she should meet. Three interests. Three numbers. No bio paragraph, no work history, no skill list. The matching is in the matching — the profile is what matters and nothing else.
Hard rules from the spec. Each one is a failure mode this design explicitly avoids.
Home, Agenda, People, Map, Profile — five tabs is five decisions every time the app opens. The conversation collapses all of them into one prompt.
8,500 names is a wall, not a network. The app suggests three faces with one reason each — that's what a guide does.
Conversation-driven doesn't mean chat-driven. Cards, faces, sessions. The only "bubble" is the AI follow-up, and it's framed as an action — not a reply.
The app already knows. Filters are the user re-asking the question. We answer once.
A skeleton row is the absolute most we show. AI complexity is hidden, not performed. If it feels slow, it feels broken.
The app's first screen is a greeting and a question. The user is already inside.