This is largely incorrect. NLHE can be more about "bluffing" because there's greater variability between ones chance of winning a hand and the potential bet size. But it still comes down to a mathematical decision, just that it's more difficult to accurately calculate the best play given the wide range of variables: your opponent's likely holdings, the bet size, the pot size, your hand.

In Fixed Limit, the bet size and pot size are all controlled and can be pre-determined, leaving fewer permutations to solve. It's easy to calculate that if a bet is a constant, then you can profitably call with holdings from XY and "higher" in a given situation, but it obviously gets harder when the bet size can vary greatly.

To answer OP, Fixed Limit is considered solved (or close to it? Haven't been following the news on this) from a mathematical and game theory perspective, and so AIs have been built to play game theory optimally. NLHE is still far, far from being solved, but there are bots that have been created to churn out small, albeit automatic, wages short-stacking (to mitigate bet-size variability) smaller stakes games where its tendencies aren't as openly exploited.

Poker is very much a mathematical game. And physical tells, reaction-time tells online are pretty insignificant in the longrun.