Inspiration

Some people can glance at a map once and never get lost again. Others — including one of our own team — can walk the same route ten times and still end up disoriented the eleventh. Standard map apps are built for drivers following a line on a screen. They assume you can stop and stare at your phone. They don't help once you look up and the destination isn't visible anymore, or once you've wandered off the "correct" path to check out something interesting.

We wanted to build for the second group: people who get turned around easily, and who currently either avoid unfamiliar places or keep their phone glued to their face the whole walk. AnchorRoute is a walking-only navigation app designed around one goal — never let the user lose their sense of "which way is the destination," even when they can't see it, even when they wander.

What it does

AnchorRoute starts by diagnosing each user's spatial cognition style (for example, whether they orient themselves by landmarks, by cardinal directions, or by large natural features like mountains and coastlines) and personalizes its guidance accordingly. From there:

  • Turn-by-turn walking navigation and a Compass mode that always points toward the destination, even off-road
  • AR mode, which overlays a persistent directional marker on the live camera view — so the destination direction is never lost, even when buildings, hills, or crowds block the line of sight
  • Automatic landmark detection plus a personal "My Landmark" registry, so guidance can reference things the user actually recognizes instead of just street names
  • Deviation-aware rerouting: if the user wanders off the route (intentionally, to explore, or by accident), AnchorRoute detects it and recalculates instead of silently assuming they're still on track
  • Spatial audio and haptic cues as a redundant, screen-free guidance channel
  • SOS location sharing for when a wrong turn becomes a safety concern
  • Walk history, points, badges, and secret missions that turn "not getting lost" into something rewarding rather than just stressful
  • A lightweight Friends feature (private invite codes) so users can compare walk stats and badges with people they know
  • An Apple Watch companion app that delivers turn cues straight to the wrist
  • AnchorRoute+, a subscription that unlocks AI-powered natural-language guidance (Google Gemini), AR mode, wider-range landmark detection, and "Temporal Bridge" — historical/era-based guidance for exploring unfamiliar areas outside the user's home region

How we built it

The app is native SwiftUI/Swift, with ARKit driving the AR overlay, Core Haptics and spatial audio for non-visual guidance, MapKit for base routing, and the Overpass API against OpenStreetMap data for supplementary landmark detection under the ODbL license. Firebase (Authentication, Cloud Firestore, Cloud Storage) handles accounts, cross-device sync, and photo storage. Google's Gemini API powers the AI natural-language guidance mode and helps interpret user-submitted landmark notes. An Apple Watch companion target uses WatchConnectivity to relay turn cues to the wrist, and a WidgetKit extension surfaces walk stats on the home screen.

Monetization runs entirely through RevenueCat. AnchorRoute+ is an auto-renewable subscription ($1.99/month with a 7-day free trial) that gates the premium guidance features (AI guidance, AR mode, wider landmark range, Temporal Bridge) while the core walking-navigation experience — turn-by-turn guidance, compass mode, landmark detection, SOS, and walk history — stays free. We used RevenueCat's Paywall components to build and iterate on the subscription screen without shipping a native paywall rebuild every time we wanted to test a new price or layout, and RevenueCat's entitlement checks gate every premium surface in the app.

Challenges we ran into

Building the trust-and-safety layer for a social feature is easy to underestimate. We localized every system permission prompt into four languages, added a dedicated Terms-of-Use/Privacy-Policy acceptance screen that now gates every sign-in method, and built a full safety loop for the Friends feature: a report mechanism, a block mechanism that instantly removes visibility in both directions, and an automatic notification to our team the moment someone is blocked — with an on-screen confirmation so the user knows it happened, not just a silent background call.

Separately, while dogfooding the app on an actual walk, we found that rerouting had quietly stopped working the way it should: once a route was set, wandering off it never triggered a new one, so a detour could turn into actually getting lost — the exact failure mode the app exists to prevent. We rebuilt the deviation-detection and rerouting logic (including handling the case where the user has stopped moving entirely) and verified the fix by walking real GPS traces through the simulator before shipping it.

Accomplishments that we're proud of

AnchorRoute is live on the App Store today, built with a real subscription business model through RevenueCat rather than a one-time download. We're proud of the trust-and-safety work behind sign-in consent, reporting, and blocking — it ended up making the product genuinely more trustworthy, not just a checkbox feature. And we're proud that we kept testing the app in the real world throughout, which is how we caught (and fixed) the rerouting bug before it reached users at scale.

What we learned

The most direction-impaired-friendly feature we shipped wasn't a clever AR trick — it was making sure the app actually reroutes when someone wanders, which sounds obvious and turned out to be the hardest thing to get reliably right. We also learned that the guardrails required for a safe, trustworthy social feature (reporting, blocking, visible confirmation) are cheap to build early and expensive to retrofit — and that building them properly made the Friends feature better, not just safer on paper.

What's next for AnchorRoute+

We want to grow the Friends feature into genuine social discovery — shared routes, collaborative exploration challenges — and expand the AI guidance mode with more personalization based on how a user actually navigates over time, without ever collecting raw location history to do it. We'd also like to bring adaptive re-engagement (gentle nudges to go explore somewhere new) driven by walking behavior rather than generic push notifications.

Built With

  • app-intents
  • arkit
  • cloud-firestore
  • combine
  • core-haptics
  • core-location
  • firebase
  • firebase-authentication
  • firebase-storage
  • google-gemini-api
  • google-sign-in
  • live-activities
  • mapkit
  • microsoft-identity-platform
  • openstreetmap
  • overpass-api
  • revenuecat
  • revenuecatui
  • swift
  • swiftui
  • watchconnectivity
  • widgetkit
  • xcode
Share this project:

Updates

Submission history