The second rule you mentioned was more in line with what I was expecting Automate to be as a crazyhouse amateur. The number of tactical options left to the piece placer are very limited, in addition to the number of squares you can place on for non-tactical reasons. It also gives more importance to which order you place your pieces in, so as to block your opponent from dropping there.
The first rule you mentioned seems interesting too, but with your first example, the rook is more expensive than the queen. That seems to impede on the viability of it, so maybe there need to be tighter rules on the range. It wouldn't happen often, but still, if the rook's cost range was something more like 4-6(rook's for 3 seems too cheap).
Either of these ideas seems sufficient to give Automate more significant strategy, I think. The first one does introduce randomness, but I think a fair amount of variant-players will be OK with 960-esque (without the occasional awkwardness of the piece placements) randomness (only at the beginning of the game, and equally), but the second would lead to more tactical and crazyhouse-like positions, which is less in line with how it currently is.
When it comes down to it, I think one of these rules is to make it more like crazyhouse, and the other to make it more like 960. Both seem good to avoid people constantly playing the top player's systems (or opponent's), as most setups are solid and it doesn't matter what your opponent does against it if you have the best setup.
As currently implemented, there is a fundamental problem with Automate. Some changes were made since its initial release, but they haven't really addressed the underlying issue.
The key problem is that there is an optimal setup, and that it isn't particularly difficult to find.
With the initial beta release, it used the original piece values. And people quickly figured out that the optimal setup involved a flood of pawns and bishops.
The piece values were changed to prevent this, but rather than solving the real problem, it just changed the optimal mix. There is still an optimal setup, it's just not the same as it used to be. It's a little harder to find than the pawns/bishops, but anyone can experiment for a few games and figure it out.
Tweaks to piece values will not solve this. Tweaks to piece values might, if perfectly balanced, result in 2-3 setups that are equally good, but no matter what you do, people will find these setups and start using them.
And once someone starts using an optimal setup, they'll generally never lose. And people farther down in the rating pool will see what they're doing and copy it. Eventually, anyone who is reasonably alert will find this optimal setup, and we'll quickly reach draw death as all the top players are using the same mix of pieces.
A couple possible solutions:
For example, you might have a game with a total of 42 points, P=1, N=4, B=3, R=7, Q=6. The optimal setup might involve a mix of bishops and queens.
Your next game might have 23 total points, P=1, N=2, B=4, R=3, Q=8. Knights would be cheap, but rooks are cheap too, and with only 23 total points the board will be fairly open, and you'd have to decide whether you prefer rooks or knights or a mix.
Each game would still have an optimal setup, but it would be different for every game, and you'd have to figure it out on the fly.
Your initial six pawns should be placed according to the current rules: 2nd and 3rd rank (5th and 6th for Black). After that, you should be able to place pieces anywhere (except that pawns can't go on the 1st and 8th ranks).
There should be a maximum of 10 pawns instead of 8. Yes, you should be able to place multiple pawns on the same file if you want.
There should be no rule against placing pieces behind the other person's pawns. These pieces may be stranded without much support, but they could also be threatening everything.
If you are unable to place a king on an unthreatened square, you lose instantly. Of course, neither side should place the king until all other pieces have been placed.
As a result, you would have tactical decisions to make during the setup phase. Right now, it's pretty straightforward to ensure that everything is sufficiently defended and all your pieces are reasonably active, so all you need to do is get the right mix of pieces. But with this rule change, where you place your pieces suddenly becomes at least as important as what you place. You can place a piece to create a threat - and your opponent will have to react to it by placing something to defend the threat.
It's not hard to imagine weird setups where, after the initial six pawns are placed, both players put all their pieces behind the other player's pawn wall.
Either of these rule tweaks would solve the problem. I actually think the two of them together would be a lot of fun: randomise the total points and piece values for every game, and also allow tactical complications during the setup phase. That would ensure that every game is unique, that we don't see all the top players using the exact same setup, and that high-level games are not all draws involving the same setups in every game.