Software architecture is only interesting if it changes what people can do. That said, technology can also move the world forward. Here is a quick look at how Topo works, for beta users who want the details.
Topo creates a new information layer on top of real cities so that you can access it from iPhone. Local news is everywhere, but it is usually buried in websites or social posts with vague location and date data. There is almost nothing that answers a simple question like “What is happening around Naka-Meguro 2-chome right now?”
Town Watch, the base layer behind Topo, collects, analyzes, classifies, and stores local news every day to observe the vitality of each area. It was originally built with coding AI to power the Topo town forecasts that we send to eight cities every day, but during that process a second system for delivering local news emerged almost naturally in conversation with the AI. I found that deviation interesting, and that became Scouter.
Town Watch scores each news item by checking questions such as “when was it published?”, “is the place within the intended TownWatch area?”, and “is the source URL correct?” Scouter receives those scored signals and the city vitality from Town Watch, then reassembles them into cards and places them back into the region.
That is what happens in the information layer. Topo, the iPhone app, only shows the result to the user. In version 1.1, we decided to try making direct contact with Scouter.
When a user launches Topo, a small signal is emitted. Nobody is told that the signal was emitted, but Scouter can receive it. Publishing a card to a place emits a stronger signal. Scouter reacts to that signal and starts fetching information for the place from Town Watch. That is what Scouter calls “observation.”
If strong signals keep coming in, the system tries to generate news for that area as a priority district. It strengthens observation on important places and, if it goes well, generates cards. Since the system’s job is to deliver information to the right place, it also celebrates the users who emit those signals as “contributors to the city.”
We also added an interface that visualizes public cards and Scouter locations on the map as heat levels. What I want to know is not only whether the software behaves correctly, but also what happens in the real city when Town Watch, Scouter, and the human signal are connected into one stage.
Whether in Japan or overseas, will Scouter truly drive and redistribute information to the right places? As a beta tester, please send a signal from your city into this urban-observation experiment that began with a conversation with AI.