Campaign · Stop 1 live
Build the tool the city actually needs.
DataRail travels by rail and stops in cities to work on their problems, not to film them. Each stop starts with capture — walking the place with wearables running — and ends with a tool built for something that stop made obvious.
The premise
A city is a problem set you have to stand in.
You cannot see what a transit system does to people's time from a dataset, and you cannot see what a street does to a wheelchair from a satellite tile. The campaign puts a person in the place with capture running, and treats what comes back as the input to a build.
Stand in it
Walk and ride the city with wearables recording. The route, the waits and the dead ends all land as positioned, timestamped record.
Find the friction
Review the capture for what actually cost time or attention — the thing residents have stopped noticing because they route around it daily.
Build for it
Ship a tool aimed at that friction, built so the parts underneath can be reused by the next city and by other people's apps.
Stop 1 · New York City
What NYC surfaced.
The first stop is live, and its capture nodes are public. These are the threads worth pulling — the map and the reels are on the stop page.
-
Transit
The gap between the schedule and the platform
Rail and subway timing is published, but what a transfer costs in real minutes on foot is not. POV capture with GPS records the difference directly.
-
Nightlife
Listings that assume you already know
The first tool out of this stop is already running: a nightlife board that keeps its own listings rather than relying on venues to publish them. datarail.org/events
-
Access
Ground truth that no map layer carries
Stairs, closed entrances, broken lifts and unlit stretches are conditions you record by being there. A wearable capture with coordinates is a way to log them at walking pace.
-
Creator
Production that has to survive the commute
Capture, edit and publish while moving, on battery, on transit connectivity. Every workflow that survives that is one a working creator can use.
The method
The capture app is the product.
The glasses record; DataRail's own capture app is what turns a recording into a node. It is written by hand in Java and Kotlin — no AI-generated code, no framework scaffolding — and developed against the Wearables Device Access Toolkit. Each reel comes back already knowing where and when it was taken, which is the difference between footage and a dataset.
Positioned by default
Every capture carries where and when it happened, which is what makes a set of reels into a dataset.
Reusable underneath
Nodes are published as plain JSON. Anything that can read a coordinate and a timestamp can build on the same record.
Open at the seams
The pipeline is documented including the part that is not finished — automatic ingest from capture to node is the current build.