Google Maps takes you to the location. Then it’s done with you.
You’re standing at the gate of something old and significant, and the app that got you there has nothing left to say. The signage is scattered. There’s no guide. That carving in the corner has no label, and you don’t know what question to ask.
JournaLens picks up where Maps stops.
I built it with Joel Dsouza over a weekend for Google DeepMind’s “Vibe Code with Gemini 3 Pro” hackathon. The idea was simple: give someone standing at a monument the one thing GPS never could: context for what they’re actually looking at.
It works in four moves.
A walking tour, generated on the spot, with checkpoints and something worth noticing at each one.

Each checkpoint opens its own chat, full context already loaded, ready to go as deep as you want.
Photograph anything you can’t name: a carving, a symbol, an architectural detail. Gemini 3 Pro identifies it, explains it, and ties it back to the building’s story.

Annotation Mode turns those photos into hand-drawn, labelled sketches that call out the motifs and features visually. The bet: some things stick when they’re drawn for you in a way they don’t when described.
At the end, it stitches the visit into a personal journal: your annotated images, your checkpoints, the highlights of what you actually understood.

The technical reason it holds together: multimodal reasoning inside a multi-agent loop. One system handling guidance, visual understanding, and storytelling, instead of three separate services bolted together.
But the part I remember most isn’t the product.
Until this weekend I’d built by proxy. Figma and evals. Engineering was a black box I handed requirements to and waited on. I’d never written code that shipped to something real. This was the first time I sat with a model and just talked a product into existence, and it kept up. As long as I was specific enough, it could manifest what I was imagining. No translation layer. No waiting.
That’s what got me hooked. Not AI in the abstract: this build, this weekend. I crossed a line I hadn’t realised I was standing in front of, and I haven’t stopped since.
Built with Joel Dsouza · Gemini 3 Pro · Google DeepMind hackathon, December 2025.