In the short history of Renegade Games, they've developed a very positive reputation, built on the backs of high-quality titles like Lanterns, Flipships, Clank, and Lotus. They're not perfect; their greatest sin is making a game called Shiba Inu House and not Doge House, which is what it is, LOOK AT THE DOG ON THE FRONT, IT'S THE FKING DOGE, WHAT ARE YOU DOING RENEGADE
Their new title is Sentient, a game about being a futuristic corporation and hiring AIs to I'm just kidding it's about dice. The theme looks very cool—the art is fantastic—but there's nothing about that theme which has anything to do with the mechanics. They could have gotten a Dora the Explorer license and turned it into a simple algebra game for kids. That's not to say the game itself is bad, but if you hoped for something that would let you sink into a futuristic world the way Pandemic throws you head-first into a toxic one, that's not what you're getting.
The game itself has each player taking control of five dice, four agents, and five assistants, in an effort to collect the cards and chevrons that will earn you more points than your opponents. Players start with a board that has a space for each of the dice and a chevron with one of the five types of AIs available. Games are three rounds; at the start of each round, players roll their dice, place them by color, and have to work with whatever numbers they got.
Each turn, a player can place one agent and any number of assistants next to an available AI card, then place the card between two of their dice. AI cards have requirements the dice on either side of them must meet in order to score that card's points at the end of the round. Making things trickier, each card has a +, -, or = sign in each upper corner; you have to add or subtract one from the appropriate die, or leave it alone for an equals sign. However, you have the option to place an assistant on one or both of those corners, which keeps the add or subtract function from happening, which you will need at some point for your dice to give you what the cards require.
(Also, once per round, players can choose to wipe the board and place their turn-order token; the earlier you do this, the later you go the following round. Turn order determines tiebreakers for winning chevrons (see below), but going later lets you set up your agents and assistants after other people are done, so there isn't a big advantage to any spot in the order.)
Five chevrons matching each of the AI types sit out on the middle of the table. Available AI cards sit between two of those chevrons (thus four cards are out at any given time). At the end of each round, players count up how many agents and assistants each of them have next to each chevron; whoever has the most next to one gets it, while whoever's in second receives a one point consolation prize. At the end of the game, each chevron is worth points equal to the number of cards of that type you've taken during the game. Thus, while one person might grab anything that scores maximum points each round regardless of type, another player can stack up three Service chevrons and five Service cards for fifteen bonus points at the game's end.
That's pretty much the game. To its credit, Sentient is well balanced between the strategies of grabbing point cards and grabbing sets. If you never collect any chevrons besides the one you start with, it'll be hard to win solely on high-value cards, but you have to be fairly unlucky not to get any unless you dump all your assistants into avoiding pluses and minuses. However, if you plan completely around good assistant placement and maximizing the chevrons you collect, and just score points on cards where you can, that's not going to be enough either—there aren't enough bonus points reasonably available to make up for a desperately low score.
The question here, as it is with other games of a similar dice-puzzle type, is how much you can deal with playing around weird strings of good or bad chance. It's entirely possible you've built an empire of military AIs and just need a couple more in the last round to solidify the victory, and they just don't come, or they fit what other people need so well they take the cards without even trying to hurt your cause. The math between your dice and the available cards might just not work one round. And this can sink you, because whatever strategy you employ at the start is going to define your game. If you have two rounds of twenty-plus points but only a couple of chevrons and a wide variety of card types, bad math in round three can't be made up for by suddenly dumping all your efforts into whatever sets might work best for you. And since you need to score a reasonable number of points each round even if you're focused on sets, a final round where you can't find anything to add to those sets isn't likely to be fixed by a random selection of high-point cards.
The randomness is a feature, not a bug, so there's nothing wrong with it. It's a simple game with enough decisions to make people go into the tank hard at times, but it's probably not something that will require twenty playthroughs to develop as optimal a strategy as is possible. This is where a strong thematic tie-in to the game itself would be nice, but as it stands, Sentient is perfectly good while not being particularly close to great.
Score: Nine busted up information bots out of twelve.