Personally, I think the hard part is finding a market for a hardware product to start with. I'm going to go ahead and assume that you've already done that and you have a validated idea, an audience that is chomping at the bit to hand over their money, and a distribution system for your eventual hardware product. That's where a lot of things fall flat, IMO. But this is an HN thread, not reality.
The first step is a system / block diagram. You need to encapsulate all the features and capabilities of the product. I mean, you won't know everything, but you should have a good idea. This will drive your prototype and your eventual bespoke hardware device. If there are pieces you don't understand whether or not you'll need them, you should go and test with your adoring audience (you have one of those, right?). You can test some of this with an early prototype, or a partial prototype.
For the prototype (really what I would call a proof of concept or POC), you should be working from off the shelf pieces, as much as possible. You're testing out an idea. Seeing your followup question, you ask about custom parts for a prototype, but you should be trying to do as much as you can without engaging manufacturing. This might involve a small custom PCB at some point, but if you can get away with soldering wires between breakout boards and a microcontroller platform, you should go that route. You really really need to validate an idea first.
Oh right, we have validated the idea...right? If not prior to this, then surely the POC exercise above. Well now you might start to think about making your first true prototype. You will be refining your idea and considering more design elements like industrial design and useability now that it's not a 3D printed case and a bundle of wires. You might engage a mechanical engineer, an industrial designer, an electrical engineer, a firmware engineer...or you might be doing that all yourself. It's up to you. But this is where things get hairy and expensive. With early prototypes, you will be making 10s of things and testing more.
Now your device is really tested (right??) and you're engaging manufacturing, and possibly going through yet another design iteration for cost down, best fit, and desiging your product so that it can be tested and you will have a high yield at the factory. You'll also need to consider regulatory testing, if your device has radios on it, and emissions testing pretty much regardless. If your product has custom plastics, you'll be enaging toolmakers and building out the entire assembly. If you have other elements aside from PCBs and plastics, you'll need to talk to wire harness manufactures, source auxiliary components, and discuss standard operating procedures (including testing) with your contract manufacturer.
Hardware is a long and arduous process. As a hardware design consultant, I have seen my share of messes personally. And if you watch the industry (or hell, just watch kickstarter), you will see people running into the same issues over and over again. Mostly that they price things poorly and they don't understand the depth of steps required to get a thing to the market. Having a well tested idea and a good audience will take you a good chunk of the way, but after that, make sure you're refining your idea many times; iteration is the name of the game.
If you're interested in just the electronics side of things, I show people how to design circuit boards and go through the block diagram -> circuit board manufacturing and testing component on my course. I show how to build a bluetooth and cellular board. If you're a true beginner, it's probably not the right fit for you. But if you've done some hardware and want to do a more complex project (or just see how it's done), this might be of interest: https://contextualelectronics.com/courses/advanced-ble-cell-...
And if the device touches the Internet, then you need to have a reality-based update program and security assessment/analysis. There is a 99.999% of a major security flaw. If all it takes is an update that gets installed automatically, good. But chances are there's a architectural flaw that will require major redesign, which means new hardware. Frankly, security has to be designed in before the hardware is designed.
The first step is a system / block diagram. You need to encapsulate all the features and capabilities of the product. I mean, you won't know everything, but you should have a good idea. This will drive your prototype and your eventual bespoke hardware device. If there are pieces you don't understand whether or not you'll need them, you should go and test with your adoring audience (you have one of those, right?). You can test some of this with an early prototype, or a partial prototype.
For the prototype (really what I would call a proof of concept or POC), you should be working from off the shelf pieces, as much as possible. You're testing out an idea. Seeing your followup question, you ask about custom parts for a prototype, but you should be trying to do as much as you can without engaging manufacturing. This might involve a small custom PCB at some point, but if you can get away with soldering wires between breakout boards and a microcontroller platform, you should go that route. You really really need to validate an idea first.
Oh right, we have validated the idea...right? If not prior to this, then surely the POC exercise above. Well now you might start to think about making your first true prototype. You will be refining your idea and considering more design elements like industrial design and useability now that it's not a 3D printed case and a bundle of wires. You might engage a mechanical engineer, an industrial designer, an electrical engineer, a firmware engineer...or you might be doing that all yourself. It's up to you. But this is where things get hairy and expensive. With early prototypes, you will be making 10s of things and testing more.
Now your device is really tested (right??) and you're engaging manufacturing, and possibly going through yet another design iteration for cost down, best fit, and desiging your product so that it can be tested and you will have a high yield at the factory. You'll also need to consider regulatory testing, if your device has radios on it, and emissions testing pretty much regardless. If your product has custom plastics, you'll be enaging toolmakers and building out the entire assembly. If you have other elements aside from PCBs and plastics, you'll need to talk to wire harness manufactures, source auxiliary components, and discuss standard operating procedures (including testing) with your contract manufacturer.
Hardware is a long and arduous process. As a hardware design consultant, I have seen my share of messes personally. And if you watch the industry (or hell, just watch kickstarter), you will see people running into the same issues over and over again. Mostly that they price things poorly and they don't understand the depth of steps required to get a thing to the market. Having a well tested idea and a good audience will take you a good chunk of the way, but after that, make sure you're refining your idea many times; iteration is the name of the game.
If you're interested in just the electronics side of things, I show people how to design circuit boards and go through the block diagram -> circuit board manufacturing and testing component on my course. I show how to build a bluetooth and cellular board. If you're a true beginner, it's probably not the right fit for you. But if you've done some hardware and want to do a more complex project (or just see how it's done), this might be of interest: https://contextualelectronics.com/courses/advanced-ble-cell-...