Where we are
Orca started in Miami. Dense urban environment, messy street-level reality, enough complexity to stress-test everything. It worked. The benchmarks validated our approach, and more importantly, people found it useful.
Now we've expanded into Europe. Multiple metros, different architectural styles, new visual vocabularies. The model is learning to generalize without forgetting what it already knows.
That's the foundation. Here's what we're building on top of it.
What's next
We're focused on three things: more cities, better reasoning, and easier access.
More cities
The immediate priority is widening the map. More European metros, selected Asian cities, strategic coverage in Latin America and the Middle East. We're not trying to be everywhere—we're trying to be useful in the cities that matter most.
Better reasoning
The model can already identify cities and explain basic cues. The next step is making those explanations feel more natural—closer to how a human geoguesser thinks through a scene.
Easier access
If you want to add Orca to your app, you shouldn't need a PhD in computer vision. Clean APIs, good documentation, simple demos.
We're not going to promise dates we can't hit. The roadmap depends on funding, technical breakthroughs, and a hundred things we can't fully control. What we can promise is that we're working on this every day.
The hard parts
Expansion isn't just adding more training data. Every new city introduces new edge cases, new architectural styles, new ways that visual cues can mislead the model. Tokyo alleys don't look like Paris boulevards. Both need to work.
There's also the tension between size and performance. It's easy to make a model better by making it bigger. Harder to keep it lightweight and still push accuracy forward.
These aren't solved problems. They're the work that comes next.
Why it matters
We believe geolocation AI shouldn't be locked away behind expensive software and niche expertise. It should be accessible to anyone with an image and a question—students, journalists, researchers, developers, regular people trying to understand the world.
That's why we're building this in the open. That's why we write about what we're doing and what we're struggling with.
The future isn't predetermined. It's shaped by the people who show up and do the work.
If you're using OceanIR, building on it, or just watching to see where it goes—thanks for being part of this. We'll keep shipping.
— The OceanIR Team