Finally someone has the guts and vision to attempt this sort of game. I had envisaged this sort of game for ages, an alternative to TW on PCs, and also tablets being largely untapped by historical RTS; CA woefully disappointing with Shogun 2: Battles. CA is good at visually and verbally abstracting what their games are meant to be about, but they continue to fail to really get to the core of the RTS gamer's deep urges.
A lot of those shortfalls have been in making the battles engaging on a macro-level; you often felt a very big disconnect between your campaign strategy and how the battle actually played out.
In most of their games they've also failed to really get a physics/particle system going that really feels visceral. The motion-capture may look 'realistic' in isolation, but it completely kills the momentum of the melee encounters in their battle and feels ultimately anti-climactic. I want a game that really packs a punch, I want the visual impact of my charge to be seen on screen; for the each fighting stat calculation to be visually represented in each stab and clash in that melee, not obfuscated by 'motion capture'. Even if it's small dots on a screen with a fixed camera zoom, as long as the particle interactions are intuitive and informative, it's going to be a lot more fun than a triple-A budget motion capture battle.
AI is largely and perennially accepted as the weakest part of Total War too, and evidently is something you've criticised and addressed a lot in your DarthMods. The problem is CA completely lack innovation or critical thinking in that area. Every year they merely tweak the thought process of a flawed-since-day-one Battle AI. What I don't think they realise is that they're not going to get an AI to feel real if the battles themselves completely lack an obvious motive behind them. An AI which has the same ultimate objective every battle (to kill your whole army) is not going to feel 'real'. Its faction/general/unit traits need to be expressed in its thinking process and the battle-map needs to actively predetermine an AI's approach to a battle. It simply does not work to build one 'chess-player' AI and then drop it into multiple scenarios and expect it to spontaneously give you a challenge and act historically at the same time. The AI needs some scriptedness to get it going on the right track before it begins to 'think for itself'. Looks like you're thinking more deeply than they have in terms of AI, looks promising.
This sort of turned into a rant about my Total War bugbears (not that I dislike the games, but I think their potential is vastly underexploited), but it was sparked by the fact that you have noticed and want to address those problems in your own game. I really do hope you make a success of this game in the places that CA has failed, because I am dying to play a TW-style RTS that does not have immersion-burying flaws like TW has.
I fear that a lot of people are going to come on this forum and, out of mere enthusiasm, pressure you down a route where they want to see complex and micromanagement intensive features in the game. I urge you to stick to your original vision; focus almost purely on making it intuitive and visceral experience. Subtly does more than complexity for these sorts of games. Good luck! I'll just be following on the forums, contributing with odd suggestions when prompted and would love to participate in a beta at some point, should you need people.