Wow - lots to see here.
I'll try to keep the answers fairly short.
Seton's is a poor map to judge the AI by. Due to it's design, LOUD is not presently adapting very well to it - he see's it (at the moment) as a purely amphibious challenge. Ideally, we want him to understand that certain start positions should focus on ground and stay out of the water. As you noted, the dual water area is also a concern on Seton's - the AI doesn't see it as two separate water areas - and cannot understand why he cannot express his naval power from one body of water to the other. For a good example of water the naval play of LOUD is like, try Coastal Conflict. We still have a lot of maps to mark for the full Naval bombardment experience to show clearly and that's one of them.
As for a no-crash mod, that's really entirely up to you. We firmly believe that it's an essential part of the game, since air engagements should not be happening over your key facilities - and all the tools are there to prevent that.
The Novax, in it's current config is simply a spy satellite - having it's costs reduced and moved down from Experimental to T3. Once the satellite is built, simply move it to where you want it to spy. Keep in mind, it is still susceptible to the anti-sat weapon that is part of any anti-nuke structure - and that's why when you turn off nukes, the anti-nukes are still in the game.
The only way to reduce the voiceovers of the experimentals is to remove them from the blueprint. Alternatively we could add a new configuration option that would allow you to turn them off, but we're not focused on that as you might imagine. As for the AI 'pings', up until now, those have really been there as more of a debug tool for me, they will be almost entirely removed in a version due out this coming week.
Artillery balance has been carefully studied for quite some time, and as such, we've had LOUD build his base to properly take advantage of not only proper shield layering (Siege Shield backed up by a regular T3 shield matrix backed by mobile shields) but a fair bit of assisting going on from his SACU and T3 engineers. It's not impregnable, but it will stand up to a considerable amount of abuse - and with artillery, timing is everything. Ideally, if shots can land within 1 second of each other, then a shield never gets a chance to got into recharge mode. Either way, we really haven't made much adjustment to shields or artillery - we've just learned how to make them interlock better.
On your AI observations, I would need to know more about your map settings, and as I said - Seton's is a poor choice on which to evaluate him. Unit cap will often lead to the situation you've described - since LOUD either can't gather enough units, or units of the proper composition, to form his platoons. It's also quite possible that he cannot calculate an attack vector with a reasonable chance of success.
As for AI and his Omni sensors, if you have the Omni cheat turned on (in game setup), then the AI will never build an Omni in his starting base unless his ACU dies - since his ACU already has that covered. He will however, construct radars in all subsequent base positions.
Join our Discord server if you'd like to share your observations with our growing community, and have some active discussions about what is working or not working.
https://discord.gg/nwR2whNGlad to hear you're enjoying the work !