As James can tell you, I’ve been vocal about this and built a tool for myself. I run into the same challenges you outlined. In a city like Los Angeles, where grids shift into organic, topography-driven streets, it is difficult to determine the best way to cover an area.
I’m working to complete all of Los Angeles and want to be as efficient as possible given the time and effort required. About a year ago, I built a web-based tool that runs locally. This was before recent advances in Claude Code / agent-based coding, so there is room to improve it for sure. The main constraint is compute, from what I understand: this is a complex logistics problem that companies like FedEx, UPS, and DHL have spent decades optimizing.
The tool outputs what you see in the screenshots. Simply, it calculates the most efficient route using two approaches, each based on a different model.
The first uses the Chinese Postman Problem to find the shortest closed path that traverses every edge, or street segment. This ensures full coverage if the goal is to walk every street and end up with a trace of the actual street plan.
The second uses the Traveling Salesman Problem to find the shortest route across a defined set of nodes. It determines the most efficient sequence to visit them, but it leaves some street segments uncovered, which I know irks a bunch of us in here (not me, so much, for the record).
The app is pretty simple to use. You draw a bounding area for your route. This can be a polygon, circle, or rectangle. You can then adjust the shape by adding points, dragging, and reshaping the bounding area, which is useful in areas like the Hollywood Hills where you may want to include or exclude specific sections.
The path is shown as a dotted line. When the route overlaps, the line appears denser, which makes it easy to track. I also tested directional arrows and a few other ux ideas to help, but prefer a more minimal interface.
Once the area is set, you can toggle between full street coverage and node-based routing. There is also an animation mode to preview the route, which helps at complex intersections and often reveals paths I would not have chosen manually.
When the route is finalized, I export it as a GPX file and load it into Strava.
I’m not sure there is a large enough audience to develop this into a full product, but I would value feedback from this group. If there is interest, I would consider pushing it further, especially if it could integrate with CityStrides down the line so everything lives in one place. At the very least, I’m happy to see I’m not the only one that fetishes over this stuff 
And fwiw, I’ve been using it when I travel. I’ve used it all over Europe and the US, and I’m currently in Brazil and its making my walking practice so much easier - now if only I could program it to avoid high-crime areas!
Dean
). How we generate the routes is incredibly nuanced/specific and is where I really need the most feedback. It’s very easy for me to build the wrong route generator.
but there’s definitely one or two processes that’s going to cover most people.