I’m struggling with overly confident players.

Why do you say they’re overconfident? Are they consistently winning fights without an opportunity cost? If so, then I would say their confidence is well placed. The fact that you jumped straight from overconfidence to character death suggests to me that you may be missing a more progressive range of costs and therefore haven’t been giving them any reasons to consider alternative solutions.

• Are enemies one-dimensional? Can the players imagine interacting with them in ways other than fighting? A powerful noble they suspect but cannot prove is breaking the law is a different kind of enemy than a slavering beast. One can easily imagine the downsides of walking into the nobles manor and slicing them in half.
• Do enemies have value beyond the loot on their corpse? Information is a common prize to dangle, find a way to talk to them (perhaps still after non-lethal combat) to gain some critical insight. Bounties for capture alive can spark some out of the box thinking, as can humanizing some enemies and introducing allies who advocate for negotiation and reform over eradication… the value of a non-lethal approach may be the favor of a powerful ally.
• Does time matter? Resting 10m to recover lost HP in the middle of a chase has consequences for the chase. Maybe we have more important things to do than murder every passerby when we’re on the clock.
• Collateral damage? Force a fight on home turf on the enemies terms. Victory could have a serious cost when the fight has been brought to you.
• Do you employ loss without death? A training-wheels consequence of underestimating danger is capture or loss of gear. There are lots of ways for a fight to go wrong that don’t result in character death.
• Do you give rewards for non-combat solutions? Ensure that solving problems outside combat earns XP/loot on par with the violent approach unless it’s a rare quest with a theme of selflessness.

Finally, consider just telling them that a fight would be dangerous and could result in death. We forget that the characters are seasoned adventurers and the players may not be. If the players lack an accurate intuition for the difficulty of a fight, let them know that their characters can judge the danger more accurately and fill them in. I did this without even a hint of an in-world justification with a first over-leveled dragon fight in a previous D&D campaign, warning the players directly that it wasn’t a fair fight and they would likely pay a serious price for rolling the dice an hoping their numbers were bigger. Because they were new players and this situation was unprecedented in their experience, I went so far as to run a round of combat in a vision-sequence to drive home how much devastation they were in for in a straight fight and then woke them up from the vision with no real world consequence for the combat other than maybe some exhaustion.This really changed their mindset and they began gathering intel and negotiating and planning how to tip the odds back in their favor through skullduggery.

In any case, I’d encourage to ask thoughtfully whether their confidence is genuinely misplaced or if you’d telegraphed to them that success is inevitable. If so, talk to them directly about a change in danger level, and start telegraphing it in multiple ways so they can see when other paths are available or advisable.

PF just isn’t built to discourage fights. Resources are easily replenished, wounds are easily cured, and consumables can patch any time limits you put in place. I’d recommend trying out some different systems with more punishing combat and more robust tools for players to avoid fights.

Another approach you can take is simply making it so a violent resolution does not lead the players to accomplish their goals as well.

Trying to get information about a big nasty with a cult, and the players decide to just murder all the cult members? Well, the players might be able to beat the cult in a fight, but not fast enough to prevent the cult from burning their sacred texts, and now you have to piece info together out of the ashes.

This is a difficult line to walk: you have to plausibly present that the outcome would have been better if they had negotiated or infiltrated, versus just “well the DM was never going to give us the text anyway”. You also have to make sure you don’t just lock off the plot because they fought.

You need a clear backup plan that’s just annoying enough to make it clear putting a little more thought into your first approach could have saved a lot of time., and maybe a slight downgrade of the end result of the plot (time is classic here, maybe a couple people the party was expecting to save got sacrificed while the party was messing around).

Are the players really overly confident or do they just think this is how the campaign is built

Meaning as long as everything is working out they will not change

There are ways to build around that with cities full of people and guards so that combat isn’t an option, using clearly higher level enemies and npcs

If the players like how things are going currently maybe the group you run for isn’t the one (I don’t want to play in a campaign where the DM limits the groups actions)

Not every fight has to end in death: have an encounter with enemies motivated to capture PCs (ransom, perhaps, or simply averse to killing), and have them do so when a PC goes down.

If it’s a TPK then they have to break out of captivity, or possibly negotiate their release in exchange for solving a problem for their captors. If only one or two PCs go down then the remaining members might have to find a way to pay the ransom, or find a way to break them out. If it’s mixed, then maybe it’s a coordinated jailbreak with PCs working together from inside and outside.

Fun scenario, but a giant pain in the butt for whatever other goals they had in the campaign, and a great wakeup to “hey, maybe I shouldn’t just be bulling into every fight”. You can steer towards a solution that doesn’t involve fighting as well, to give them a forced crash course in their characters’ nonviolent capabilities.

Talk to them, because if they like this diablo style of play, you might scare them off. But maybe you can put the elite template on a few enemies to make it harder.

Otherwise see what the enemies have or if there are enemies with cool features. Swallow is always fun, drained is super annoying. Find enemies that pick on the weaknesses of your group. Low will? Get a scare aura on your enemies and drop the fleeing condition on the front fighter.

Some possibilities:

I had this problem with PCs that antagonized everything in the world whatsoever and got no allies. I realized it was directly correlated with the perception that, although I was not railroading them, I was GMing an adventure in railroad format (read: 99% of adventures in the market made after the release of dnd 3.5). In railroad format, the encounters are carefully balanced to always be winnable. The PCs act brazen simply because they know they’ll get away with it every time. Even if the consequences are nonphysical (you fought the Prince of Mendev, now the guards are hunting you), they know they’ll brush those consequences easily as well because the GM’s plot must continue. Which brings me to…

Railroad is the ugly word. No one will ever admit they are railroading and everyone has a different version of what “railroad” means. I recommend some reading of The Alexandrian blog and bankuei’s blog. They have the pretty radical definition of raiload. For both, railroad is when you run a prewritten plot. Alexander writes this in a very gentle manner like a pre-school teacher, Bankuei will accuse you to your face like a drill instructor.

I don’t agree with their definition of railroad, but I recommend you to read both nevertheless. I had this same problem, as I said before, and this was the solution: Stop running pre-written plots, as they were the reason for 80% of my woes in tables (the other 20% are dealing with lost players who don’t know how to play without being railroaded and players who think the GM is their fucking personal jester instead of another player with their own opinions and boundaries).

Fatigued maybe? Requires a full night’s rest to clear, so it would impact the rest of the adventuring day.

My current group were once pretty overweening about combat, because they were aware I wasn’t altogether familiar with Pathfinder and wanted to avoid an accidental TPK.

Then I threw a bunch of rust monsters at them. Now, I only have to mention that, say, a door handle looks a bit corroded, and they run for the hills.

There are losses other than death that allow you to inspire caution in your players, and it’s perfectly fair to use them. Loss of valued property due to poor decisions is one. It doesn’t have to be rust monsters, it could be getting arrested and weapons being confiscated until a heavy fine is paid, for example. Gold doesn’t automatically replenish after a long rest.