Startups are extremely risky. They're a terrible way to make a living, and they're an even worse way to get rich. For the vast, vast majority of people, I'd suggest not starting a startup at all. It's difficult, painful, tedious, and frustrating. It requires a huge amount of sacrifice.
If you're determined to do it, the book Lean Startup lays out a good methodology for actually building your product and minimizing the cost of failure. It's not exactly a "start here" kind of book, though.
There's a series that's been on HN recently called "How to Start a Startup" that you might want to check out.
Regarding your question, "What if it goes to hell, and I'm left with nothing?" Chances are EXCELLENT that it will go to hell. If you can accept bankruptcy, that might not be an issue for you.
If you have a day job that gives you lots of flexibility and extra time, you may be able to bootstrap something on the side. You'll need access to customers for quick revenue. A good rule of thumb is that, if you can't build it in a weekend and sell it to someone, you should look for something simpler and more needed by your target market.
Note that when I say "access to customers", I mean intimate knowledge of those customers as well as personal connections. I've never worked on construction, so I wouldn't build a SaaS for construction foremen. Who are the smartest people in your network who would instantly understand the value of a new software product? Do they have a lot of money? Are they in industries where they adopt new technology quickly?
To answer your question about where to begin: find a group that you know that struggles from a problem. Ask them about it. Then spend a weekend building something that solves their problem. Then ask them to pay for it.