I just realized that I have been a member for almost 10 years now. I have seen the game evolve, I have learn how to succeed and experiment using different business models. I am looking forward to 2026, and can’t wait to see what it’s going to bring.
A few days ago I came up with this crazy idea of using Notebook LM to experiment and see how AI could suggest improvements in the network of one of my little airlines.
The result is very interesting but the main difficulty was to add sources. By sources, I mean, current flight plan, airport database, aircraft profiles. Some sources were found easily but difficult to import as sources for Notebook LM.
Flight plan was found on ASMR, which has an option to export it
Aircraft profiles. I haven’t found any database yet, but AI seems to have an accurate idea of their basic capacities.
Airport database. I started copying and pasting from AWS. That solution is ok because my airline has only 15 stations, but definitely not viable with an airline that has a lot more stations. I found a database, but it doesn’t includes MCT or traffic level. It would probably do it for now, but I am looking for something more accurate and detailed for the game.
I wonder if someone in the community could help me pull out a database. My idea is to release the results of that experimentation to the community. I see that as a tool to complement the ones already available. More like a way to optimize aircraft use and waves planning strategy. It would be able to keep track of what you changed, and what worked and didn’t.
Who am I to judge how players play the game. But I can’t help but think: If you let AI play the game instead of “YI” (Your intelligence), why bother playing at all? Isn’t it sort of the point to figure stuff out on your own? Or am I just old?
Fair enough. A bit of experiment is what makes the game more interesting instead of looking elsewhere after all (I believe). ASX, ASE as well as ASRM tools were built to bring more depth to the game and keep playing. The idea is the same here: Being able to analyze the amount of data we have available.
Unfortunately, I personally don’t have the availability to spend 10 hours a day on the game optimizing, like some of my fellow competitors who manage to grow an airline up to 500 aircraft within a month, so I figured using modern tools would be helpful. The best way I believe to fight AI (if you’re against it) is to actually try to understand what it can and can’t do.
Notebook LM is not perfect, it will probably never understand the concept of pax distribution in this game, and I am not looking for that. We know that the obvious routes are not necessarily the most profitable, as sometimes the game is not able to replicate what’s the traffic is like in real life.
But what Notebook LM can do, is suggest things to get started, then the game will decides and we can go from there.
If you believe this thread as nothing to do with AWS, that’s fine to me, I am just looking to experiment here, I won’t be offended if you delete it.
This is one of the main reasons people requested an API. It allows the players to view the data in ways different from the way the developer envisioned. Different reports, different ways to analyze the data, etc.
I don’t know about submitting to AI (I’m not a big fan of that either), but just being able to see the data, including historical data, in different ways can help.
Yeah I agree. I was reading yesterday about the api and had the same thought.
Having a bit more history would help us for sure. Right now the financial statement only showing 3 weeks is a bit short.
Also not being able to see the transfer data at our stations is a big missing feature. Especially because a big part of the game is the ability for our passengers to transfer between flights. When I sign an interline agreement, I know it could be beneficial but I never know the scale of it.
Back to the AI thing, I don’t think it would be able to access the API. Also Notebook LM work with static sources that you add.
Example, I imported the timetable of one of my airlines (from asrm) and it was able to read it and understand it and eventually recommend some tweaks.
It was lacking the load factor, so I added a screenshot of my stations statistics which gave it an idea of the load factors. It was then able to understand where I was over/under capacity (something I knew already but It wasn’t the point).
I asked it to rework the whole timetable with 3 waves and strategically scheduled the flights for optimization. The result is quite interesting. At the beginning it didn’t included any overnight stops at outer stations to feed the first wave (simply didn’t get the idea of doing it). When I asked it to do it, it then worked on it. But something else was missing. Airport curfews… Another wave of screenshots and it re-did everything. Slot availability was the next issue.
Now, all of that is just testing and intellectual work. AI will try to respond to what you ask, so the thing is to find what to ask and how.
As you know the servers evolve daily and what’s true today with the slots won’t be tomorrow. So this might not be a long term solution, especially with big airlines, but it’s nice experimenting.
