Wander Spot

Inspiration

We've all gotten lost before. And it's not great. Not in a dramatic way — just that quiet, disorienting feeling when you turn a corner in a new place and realize you don’t know the story of where you’re standing. Maps tell you where you are. Review apps tell you what’s nearby. News feeds tell you isolated events. But none of them tell you what a place means. We wanted to build something that didn’t just retrieve data, but interpreted it. A storytelling lens for the physical world.

What it does

WanderSpot is a storytelling AI that turns any street into a narrative experience. You upload a photo. We ask for permission to access your location (via permission). You confirm. Then you choose: Street Story → A narrative overview of the area Local Story → Structured local updates

How we built it

WanderSpot was built as a modular pipeline that transforms raw geographic signals into structured narrative output.

We used Claude Code to rapidly prototype and iterate on both backend logic and storytelling prompt design. Rather than treating AI as a black box, we designed controlled prompts that only allowed specific, verifiable signals to influence the final narrative.

The technical flow includes: Image upload & location extraction Location is achieved through user-granted coarse location. We also used google geocoding APIsto derive neighborhood-level context to ensure narratives are in the correct geographic scope Google Maps API for map rendering Local news APIs to pull structured signals about recent news updates Claude API for narrative synthesis, transforming structured inputs into coherent, readable stories The key architectural decision was separating signal aggregation from narrative generation: Structured Local Signals → Controlled AI Story Synthesis

Challenges we ran into

The biggest challenge wasn’t technical. It was defining scope. “Tell the story of a place” is an enormous idea. We could have added crime dashboards, real-time alerts, predictive analytics, personalization, and social layers. But we realized that trying to do everything would dilute the core concept.

We narrowed the problem to one question: How can AI interpret structured local signals into a clear, meaningful narrative? By focusing only on location confirmation and storytelling synthesis, we kept the MVP cohesive and intentional.

Accomplishments that we're proud of

Turning a broad concept into a focused, working MVP
Designing a dual-mode system (Story Mode + Local Mode) from one backend pipeline
Building a human-in-the-loop location system to simplify complexity
Creating narratives that feel grounded rather than hallucinated

What we learned

As beginner coders, we learned that building something meaningful isn’t about complexity. It’s about clarity. The hardest part wasn’t integrating APIs or using AI. It was defining scope, structuring inputs, and deciding what we want to create. We realized that constraints improve output, and thoughtful design matters more than adding features, especially when timing is tight. We also learned that asking for help isn’t embarrassing. It’s part of building. Iterating, debugging, and refining with guidance made the project stronger. Most importantly, we learned to build something we genuinely found interesting, not just what we thought would look impressive. Creating for ourselves made the product more intentional and authentic.

What's next for Wander Spot

Next, we want to deepen the storytelling.

That includes adding “Then vs. Now” comparisons to show how neighborhoods have evolved, improving signal quality for smaller cities, and refining perspective modes, like food-focused, history-focused, or culture-focused narratives.

We’re also interested in exploring longer-term change detection: identifying areas that are gradually shifting rather than just reporting recent events.

Ultimately, we want WanderSpot to become a storytelling layer for the physical world, helping people interpret place, not just navigate it.

Every street has a story. WanderSpot tells it.