Live crowd, music, cover, and line updates help you find the right place in real time.
OUTBND shows you what is happening at nearby venues right now. Every update shows when it was reported, so you always know how current it is.
02 / Signal Architecture
Verified venue facts supply durable ground truth. Live room state flows through three real-time reporting channels: the people running the room, the crowd inside it, or ambient edge sensing.
03 / Live Simulator
Explore a fixed Houston night, hear what a room sounds like, and follow the same map-first decision path as the current iOS app. Venue status and the highlighted walking route are simulated.
6 current reports
1 min ago
Live status · simulated 1 minute ago
04 / The Trust Model
Every live claim carries a source class and an exact timestamp. Updates move from Live to Recent as they age, making their freshness clear at a glance. Crowd level, cover, and music updates are always paired with the time they were reported.
High confidence. Displayed with full hue prominence and active radar radiation.
05 / SAMI
SAMI is the ambient signal layer in OUTBND. It contributes confidence-weighted room-state input between human updates and keeps the system honest when direct reporting pauses.
SAMI supports venue-state continuity as a supplemental feed, not an unverified authority. Its contribution is always source-attributed and interpreted through the same decay rules as every other live claim.
SAMI supplies low-friction ambient signals for room-state confidence without requiring customer interaction.
SAMI readings are fused as one channel in the trust model and never override verified venue facts or direct updates.
If signal quality drops, OUTBND clearly lowers the confidence of the reading.
If SAMI confidence drops or signal continuity is interrupted, OUTBND lowers the confidence weighting and marks the signal as unavailable rather than displaying false certainty.
06 / Technical Architecture
The stack is arranged around one question: How fast can a phone discover what is happening nearby right now, and how confident should it be in the answer?
SwiftUI · Mapbox Vector Maps · MapKit
Engineered for one-handed operation in low-light environments. Custom nocturnal vector tiles minimize battery drain, while MapKit calculates precise turn-by-turn walking times.
Fly.io Dual-Node Dedicated Compute
Maintains persistent WebSocket connections with single-digit millisecond latency. Always-on compute nodes eliminate cold starts and broadcast venue status immediately.
Supabase Dedicated Database + RLS
Executes radial geospatial queries (nearest active venues within 3 km) natively in the database engine, avoiding client-side geographic overhead and memory pressure.
In-Memory Pub/Sub Event Broker
Pushes live room-state changes directly to active map screens. Connected devices update row states in place with zero background polling and zero interface stutter.
Automated Swift Model Codegen
Swift client models are generated directly from the backend schema during build. Client and server contracts remain synchronized with zero manual decoding maintenance.
Cloudflare Mail Auth + Salted Tokens
Pre-provisioned door and venue operators activate through secure, single-use cryptographic tokens, preventing unauthorized broadcasts while keeping general usage passwordless.