The Dublin Marathon map nobody had made
Marcelo Varge · 2 November 2026
On the Saturday before the Dublin Marathon I was trying to plan my Sunday. I had to vote in the Brazilian elections which were from 8am to 5pm and then get out of Dublin with roads closing at different times all over the city.
Disruption information wasn't easy to find, I saw friends asking around about the closures, and the official site had a long list of streets and times and a PDF with an image of the route. There was no map where you could just check your street.

So I gave the PDF to my coding agent around 10:30 and asked for a simple map. The first version was live around 11:20.
It's Vite and MapLibre with the street names matched against OpenStreetMap, running on Vercel. The agent opens a PR for each change and I merge it. I merged most of them without much checking in fairness, but I wanted it out before Sunday.

The AI got a few things wrong and I had to steer it. Someone told me the route was going along the canal towpath near Rialto and that was fixed in about 10 minutes. Then someone on Reddit pointed out the finish was on the wrong street. The line is still about 800m longer than a marathon, so it's only approximate and for now you have to check what it gives you.

Once live, the limits took more of my time than the map. I was on the free Vercel plan and if it went over then my other sites would be paused too. I worry about this stuff so the caching was there from the first version and the agent wrote a script to check usage every 5 minutes. I also had a backup copy on Cloudflare in case I needed to switch, it wasn't fully ready but it was something. On Sunday, while I was out, another agent was watching it and would ping me if anything went wrong.
It got around 12 thousand visits over the weekend, about half on Sunday with the peak on race morning. Before AI I wouldn't have bothered building something that's only useful for one weekend, it wasn't worth the time but now it was actually fun to go after it.

The data was public and it cost me around $30 in tokens, not exactly free but surely low cost for the gains, so I was a bit surprised nobody had done it. I think there's an invisible barrier there, AI wrote most of the code but I had to know what to ask for and spot when it got things wrong. I guess most people just scrolled the list and didn't see it as something they could build.
I don't think it stays like this for long. Probably soon you open that PDF and something already suggests the map, or it gets easy enough that you don't need to know what a PR is or worry about limits. And I'm looking forward to whatever we get to do once that part is gone.
The map is still up at marathon.mvarge.com and the code is on GitHub if you want to have a look.
I used AI to help write this post.