It's time for a
Spot on! (#1)
I decided to make a few screenshots of individual AIs and write a few lines about them.
Note that they are choosen randomly, not by my preference or anything. Please don't hold a grudge against me if you're an AI author and your AI has not been covered by a spot on (yet

).
The first "spot on" is a double feature

presenting a veteran of OpenTTD AIs, as well as a newcomer!
I hereby present:
AdmiralAI and Denver & Rio Grande
(AdmiralAI: yellow; Denver & Rio Grande: purple)
AdmiralAI is an AI by Yexo, one of the first AIs available (june 2008), and was the first AI with train support (after it had become possible in the API). Yexo is not only the author of this AI, but has also become developer of OpenTTD and is responsible for the NoAI framework.
AdmiralAI supports trains, road vehicles and planes.
Regarding railway transportation, AdmiralAI tries to reuse its existing tracks. It will start with connections between two points, but if there's existing track somewhere nearby and going in the right direction, it will hook its new route up to the existing one and thus start to build up a "network".
In the screenshot, you can see this principle in action. Early in the game, AdmiralAI built a coal line to this power plant. Later on, it decided to build a second coal line ending at this location, and instead of building a seperate station for the new line, AdmiralAI hooks up the new route to the existing one right before the station and thus reuses its infrastructure.
This is a feature, to my knowledge, yet unique among all OpenTTD AIs.
--
Denver & Rio Grande is a brand new AI by Dustin. It is still under development, only Release Candidates are available yet. The first public version was made available only this month (July 11th)!
Denver & Rio Grande is a freight train-only AI.
It builds no network, only point-to-point connections that are single track with passing lanes to allow multiple trains on them (kind of like the original AI). To construct its tracks, unlike most AIs that use the library or a modified A* pathfinder, it has a very different approach with the aim of making pathfinding much faster. The main pathfinder will just try to take the most direct path, and if it hits an obstacle, it calls different sub-pathfinders specialized in finding a way around the obstacle.
Unfortunately I found that the pathfinder, while doing quite well on "natural" obstacles (towns, industries, lakes) does not behave very well with existing tracks as an obstacle. It has only very slim chances of going over them, then it will try to find a way around them, which will most probably fail. This is an explanation for Denver & Rio Grande not doing very well in this battle royale, where there are tracks all over the place

As this AI is still under development, I hope the author further improves it and includes a sub-pathfinder able to cope with existing track better

looking forward to seeing a more competitive version of this AI in the next battle royale
That's it for now, see you next time with a new AI at "spot on!"

---
PS: @Dustin, your post came inbetween - hope my post answers some of your questions. Yes, your AI is alive and kicking, it even managed to build a new line just recently (at the very edge of the map where no competitors' tracks were in the way).