Not sure how exactly this can be implemented, but I would probably introduce some rule where each side will need to commit to their next move before getting to see their opponent’s last move. A crude example - white makes move 1, and commits to move 2 before he gets to see black’s response to move 1. Similarly, black will commit to their own move 1 before getting to see White’s opening move 1. And when Black will get to see White’s opening move, he will need to commit to his move 2 before he gets to see White’s move 2. Probably butchered up the description, but that roughly describes it.
Obviously this can run into situations where a pre-committed move might be illegal - and I have not thought about how to handle that yet - but this pre-commitment rule may challenge the classical notion of chess being a perfect-information game.
I know some “fog of war” variants tried to introduce a stochastic factor, but only spatially (i.e. hiding a portion of the board). In my variant, technically the whole board would be hidden before one side pre-commits their move. This will be a test of thinking ahead, as well as how well does one know their opponent (researching one’s opponent is an important factor in top-level chess as we all know, this variant just makes it unavoidable).