So you have a few major advantages with sand-boxing in Pathfinder 2e, I'm a GM who does only homebrew (with a very little bit of PFS during a con being the only exception) and I've been using the system just about weekly since launch.
1. Exploration mode provides a super good built in procedure for sandbox play, allowing you to crawl large dungeons and areas of wilderness. It sets expectations for tracking time and codifies a system of activities for your players to use (without precluding additional ones) while they move through the environment in minimum increments of 10 minutes. It functions like a marching order where each member of the party gets to take on roles, ranging from searching for traps and secrets, watching out for danger, studying the environment for lore, sneaking, and so forth. It creates a more elaborate game play space than other modern games does with exploration.
2. The encounter guidelines function with a high degree of accuracy, there's no real need to push the players through x number of encounters to exhaust their resources, severe fights are hard even if the party is fully rested (though a prepared party going nova can still mitigate this to an extent.) This means giving the party control of the pacing of their adventures doesn't have major problems. It also provides a relatively simple gauge for how difficult an encounter will actually be, allowing you build them in a modular manner, and track how hard they'll be if they combine/split apart. Adventures and adventuring areas are actually really easy to create, and the monsters in this game are really fun in their own right because they have cool abilities, you also have an ever expanding supply of cool monsters, traps, and magic items to design with.
3. Crafting doesn't break the game open, though its usefulness is game dependent-- its more useful in my game where access to high level settlements, or high level magic items in low level settlements, isn't a given. Magic Items similarly won't break the game open, and with the variant rule, you could easily run high treasure, or low treasure without any problems. Coming from 5e, a flaming sword here is much less problematic for your player to have, which means they feel freer to actually have it. This will make your sandbox much less breakable regardless of what your players do.
4. Downtime mode is nicely fleshed out, letting you engage in activities that earn income, retrain your abilities, and do other stuff (some of which is class or archetype specific) when you're not adventuring, in a well balanced system. This is important in a sandbox where the players might decide to pursue their own goals in town, either waiting for something specific to happen, or to try and gather resources and prepare for their next adventure. There are rules for settlement stat blocks that can flesh out the game world, and subsystems for everything from infiltration to research, they're all based off a mechanic called 'victory points' that make them easy to learn, but each is a little different.
5. Experience is a flat curve, and the game sets out different levels of 'accomplishment' exp that should make it easy to track their progress without having to constantly adjust your amounts to keep up with an inflating target, while still rewarding them when they aren't fighting.
6. Finally, the large array of options, common and uncommon seem nice for a sandbox world because players can be more invested in their own growth. Rarity allows you to designate some things off limits because they wouldn't mesh with your game style, or even just ot hold them back as rewards for engaging with the world-- e.g. to get a certain archetype you have to join a certain organization, or whatever. I'm considering removing bags of holding in an upcoming pirate west marches because I want transporting the treasure, in general, and in safety, to be a present theme, as they plan their voyages to their destination and back to a friendly port on a hexmap.
Then you have one kind of big problem:
1. Level is added to most of the numbers, things more than 4 levels above, or lower than 4 levels below, are numerically outside your players strike zone (the things below will go down like wet paper and not have any real chance of hitting, the things above will nuke them quite unfairly) and those encounter guidelines I mentioned are heavily based on the level of the monsters relative to the party. This does create a scenario where areas of your world will be straight up too dangerous to go to until they level up, or will be too weak to bother with. Depending on your approach this could actually be a good thing, if its well communicated the players can set long term goals to revisit an area or something and make the world feel more alive as it isn't scaling with them (they'll feel growth because it'll let them do things they literally couldn't before) but some people dislike there being any limitation on 'we want to go here and explore this.'
