Longtime Netflix CEO Reed Hastings is renowned for building a video streaming service that has transformed entertainment, but it probably wouldn't have happened if not for his friendship with serial entrepreneur Marc Randolph.

While brainstorming with Hastings in 1997, Randolph conceived the DVD-by-mail service that launched Netflix. He then directed Netflix as its first CEO before handing off the reigns to Hastings in 1999.

Rather than retire and live on his Netflix fortune when he left the company in 2003, Randolph decided to counsel early-stage startups and their founders. He also worked as part-time executive at data analytics startup Looker, which Google bought for $2.6 billion in 2019. He also wrote a memoir/advice guide, "That Will Never Work," and hosts a weekly podcast.

Randolph, 64, recently shared his insights with the The Associated Press at a cafe near the Santa Cruz, California, post office where he mailed the first test disc for Netflix's DVD service in 1997.

What have you learned counseling startups?

You expect you are going to be helping with a go-to-market strategy and the technology, but a huge piece of it is marriage counseling. For a lot of the problems you face as a CEO, there is no one else to talk to about it. So if they are struggling with something, they can't always go to their team and talk. A lot of times they can't really go to the board either, because they don't want them to know they are struggling or the board doesn't really understand the nuances of the problem. And they can't go to their friends because their friends don't know the details. There really is no one else that is impartial and understanding and knows enough context. So that ends up being the most gratifying and useful thing. I don't build companies; I build CEOs and founders.

What's your view of the current economic uncertainty and how it's affecting the tech sector?

It has been a remarkable decade for tech, accentuated by the last two or three years, but there are a lot of cyclical things that are going to be corrected. Certainly, one is tech valuations and expectations. You had this crazy period for the past 18 to 24 months where employees called the shots, saying, "I don't like what I see, I can get 50 other job offers tomorrow." That is going to correct, and I think it's going to correct in a positive way.

How did the dot-com bust in 2000 and 2001 affect Netflix?

It was a powerful lesson. We were going to be a [entertainment] portal. Had we gone that way, we would have been defunct because the model was unsustainable. We were lucky it corrected when it did. As painful as that was, it forced upon us this discipline where you realize that whatever service you provide had to cost you less than what you charge for it. What a concept! Some startups are starting to realize, 'Holy [crap], this is all about cash flow. All that talk about growth, forget about it, we have to focus on getting through this.' Those that don't are just going to keep accelerating right off a cliff.

What's your take on the type of tech hyperbole that resulted in Elizabeth Holmes being convicted for fraud?

A mistake would be to think it's a binary thing. There is a point where it verges into outright dishonesty. And even worse, dishonesty where it's going to hurt people. Is it too far for Tesla to claim heightened self-driving? Yeah, especially if someone thinks they can just go to sleep in the back seat in a car that clear isn't autonomously driving. But for the most part, most people don't go too far.

What I don't want to do is throw the baby out with the bathwater here. Are there some extremes? Absolutely. But is it appropriate to talk about the vision for a product before it's completely ready? Absolutely, at the risk of it sounding like hype. Telling someone 'This is what I think this could be' isn't the same as telling them 'I guarantee this is what this will be.' It's a really interesting question because I have the superpower to talk with tremendous convictions about things that I am not entirely sure about. And I have to be really careful about that.

Michael Liedtke reports for the Associated Press.