It is generally a problem with strategy games - the difficulty curve is inverted from what it should be. The hardest part of the game is the very start when you're smallest and least prepared and the game gets easier from there. In every other genre of game, things start off easy and get harder.
I haven't seen a game that really solves this problem (more importantly, solves it while remaining fun).
I mean I used to play Medieval 2 Total war and aided in making a very heavily modded game, also with focus on improving the AI to a large degree.
Some solutions include programming the AI to take into account when the player is at war and with how many enemies as a decision factor kinda like how CK3 handles factions. In other words the more enemies you have the more likely other AIs were likely to pick the player as the target for territory grab or invasion. So the larger you got...the more enemies you'd touch...the more likely you fight multiple wars on multiple fronts - and becuase the AIs would sorta band together on the back/programing end, not by formal in-game alliances, but more by informal attacks of opportunity elevated by script.
Also heavily modded the AI to prefer naval invasions so you'd have people from halfway across the map coming and randomly landing on your shores to raid you just when you think you have your current conflicts under control - instead of the AI just chilling on Malta or Iceland doing nut-all.
That and having naval ships and allowing the AI to win a fleet battle, sink some of your ships - and with it some of you army - could catastrophically end your summer campaign before it even begins. It also punished the player disproportionately more, more often of the time - becuase as the player advances faster than the AI the player building more expensive advanced troops would lose a fair portion of said advanced or experienced troops, hurting elite armies more, allowing the weaker AI troops/armies to then be able to better fend off now half size elite army of the player. The ship battles were all auto-resolved back in those days, so one unlucky naval battle you could see half or more of your elite men-at-arms sinking to the bottom of the sea after years of building and preparation for your next land grab. Also small/nations like Malta or Venice, the AI would build more naval ships than army so they could just keep sinking your fleets and forcing retreats of your navy...not allowing you army to actually land on their shores. This definately affected and slowed the player down becuase if you didn't win a decisive naval battle or invest enough into ships...your Uber land army was worth nothing if it couldn't land and fight.
The modders and I did in my humble opinion a fantastic job of greatly improving the game, ai, and difficulty and fun factor and variation of enemies without implementing "cheats".
Balancing how many enemies you pick a fight with was a core balancing factor and slowed the players map painting a fair bit. Still wasn't impossible but instead of besting the game in 40 years...it would take 300+ years to map an empire and even then it could swing to game over fighting 20 small nations at once. You could have the troop count but not the stacks to fight all over the map. All the while nations you eliminated reappeared a hundred years later and God help you if they spawned in place of normal rebels and captured one of your vulnerable core heartland castles with your upgraded barracks that allowed you to build top tier troops - becuase basically...you the player just funded their tiered up settlement they now are going to absolutely use against you and recruit their top tier best of the best faction troops to unleash havoc in the middle of your empire. Trust me - all your foreign expansions wars and border skirmishes suddenly are of no import.
Also, and what I am refering to above, a key feature we programmed was reviving nations/cultures. Instead of random rebels or popular uprisings/revolts in poorly managed cities spawning generic slaves or rebel armies...occationally if a faction / nation was previously eliminated...it would spawn that factions army instead on the map...reviving say a conquered Wales...with an NPC "Prince" with a superior scripted army to regular rebels. This posed a VERY significant threat as having higher tier army of "loyalist" faction rebels with a declared leader could easily capture a city or castle in your heartland while your standing armies are out on your border lands and then all of a sudden that eliminated faction was back on the map again building their unique cultural troops right near your previously "safe" inner empire. Of which you would have the same diplomacy options as a full and regular nation and all the problems therein.
This kept the game remarkably fresh and was one of my favorite features...as eliminating factions / nations wasn't a perminant thing.
This was emphasized and facilitated by the fact that corruption and discontent or whatever it was called increased in penalty the further from the capitol the province was lead to more frequent revolts/independence the farther from the capitol that province was...especially after death during succession. Kinda like CK3's option to have disconnected baronies or counties become independent on succession.
in summary - had some really amazing games and late game was always as challenging (though in different ways) and exciting as the early and mid-game.
Honestly there are so many ideas to make the late game better.
Another thing that was different is that you could destroy building when you captured a province, get some $$ from raising it...then just give the city back with your diplomat in exchange for a truce or peace. This was a great strategy for slowing down a blobbing or powerful AI or player becuase it would set their military back a step but only in those provinces. So you didn't have to manage revolts etc. or just let the city revolt and become independent. But you can't offer to exchange baronies in CK3 through diplomacy so...=/
Another option was to sack the city and slow development by damaging building that needed to be repaired before being used again and giving gold similar to CK3's raiding mechanic...which the AI would do. The goal wasn't to hold the city/castle long term, just capture it, loot it, and leave. Even if the player/AI immediate captured the city back (like even if the city revolted against the new owner and went independent first)...the economy would definately suffer...so this kept the late game interesting as you could either focus on spending money building troops and armies and fighting elsewhere (expanding) or trying to keep your heartland cities in a viable economic state or repair.
But CK3 lacks a lot of these options despite some cities being raised to the ground IRL.
An idea to maybe consider would be to maybe to see an option to raise a barony when captured and reduce all building levels by 1 level or a random amount of buildings to represent the devastation cuased each time it is captured/sacked...and if a settlement is captured or changes hands often enough and no building(s) remain, the final sacking would destroy the main building (city/castle/temple) and remove it from the map entirely - to be re-founded again at a later date...like at the start of the game when baronies are empty.
On that note a decision to change castles to cities/temples or vice versa would be nice too. I really hate when I want to station archers in a forest area only for the AI to have build a temple there first. =/
I'd also like to see the basic main vanilla starting cultures/kingdoms/Empires decisions to recreate them if destroyed to be a driving AI goal. This would push the game towards more historic norms...while still leaving room for shenanigans and ahistoric diversions.
I get this is a character focused game but I think characters should have goals beyond seeing how many times they can cheat on their spouse without them or the Pope finding out. =>