Hello DGM
A fair question, hope we may give some inspiration. The key concept is increased fun through self-discipline. Shackle the economy, as was it a wild beast, and avoid excess.
There is a reason why early game is considered most interesting and fun strategically in about every TW game that ever existed. Because you're yet not so large, so rich and your armies so many that no opposition is of concern, and the game is not yet merely reduced to industrial scale grinding of enemy stacks. DCI has been designed with the aim to secure that this challenge is long term, as fars as we may, without being crippling.
Economy is the key to this, for the player and for the AI.
If you can afford five stacks for two fronts, or have four stacks when the enemy has three, challenge is in question and unlikely. To offer challenge the enemy then won't have only three stacks if you have four though, will it? The AI will have ten, and more coming! But it do not have to end up with no concern for money and grind.
For the player you must make sure that the costs for building is relevant and bonuses don't become so many or so large that money becomes of no concern. In important symbiosis to this is that the costs for holding an army (costs for recruiting but more importantly upkeep) should be significant and balance out: when you field armies you should not be rich (have some money unless you spent ill, but not rich).
The reason for this is that the game will be thrilling when you have one stack but two fronts to protect, or have two stacks but the enemy has four. And if this is fine-tuned then it can scale well, so that the players have to consider their economy and how much forces they can afford to send here or there still at turn 200.
This require tests and rebalancing, certainly, but I'd argue it's worth it (as Earl refer to above, hehe).
For the AI you must think the same way. If you have scripts that give money bonuses to AI factions, keep them limited. However, since you already limited the players economical capacity, it is resonable to decrease unit replenishment rates for all factions so the AI can not misuse the economical edge that it do has, it will not stack spam but still offer a challenge and you can limit garrison scripts and spawns in general.
On the plus side, it makes battles important throughout a campaign. If you and the enemy have few armies, defeating one or getting defeated can suddenly have great ramafications. And with great ramafications comes great tension
This as noted require self-disciplin from us modders. We as a breed seems conditioned to think that "if we can add something we better should", hehe. More of something is always the solution, less of anything hardly comes to mind.
So we might unintentionally boost the problems of low difficulty + grind, instead of designing against it: it is very common in general as modders know we are free to add what's instantly fun (to make for us or from player feedback), resulting in excess and inflation.
Say that we toss in buildings that offer bonuses and money, for the short term fun of instant rewards, because players love things to build and troops to recruit.
Then we reach the situation when the enemy have to great forces to handle. Compensate that with more income and bonuses so you can get more troops! But now its to easy! Boost the AI's ability to muster even more forces! And the circle begins again. The spam grows, while the design do not try to amend the root issue, but makes the situation worse.
When the fun of "getting new stuff" is the point and guiding light at the expense of challenge; then balance is compensated for with the spam, instead of sufficiently designed for.
This is a view point, naturally, I do not claim the have 'the truth', but low spam and long term entertainment is what we aim to offer by this philosophy. Hope folks will enjoy it