As I understand Newhouse: All the Glitter, Power, & Glory of America’s Richest Media Empire & the Secretive Man Behind It, its author, Thomas Maier, said the Newhouse family was in the business of providing information to Americans. They owned magazines, newspapers, and cable systems, but newspapers were the empire's backbone. Growth potential was capped, but profit margins and cash flow were fat.

For many years, newspapers were the best way to distribute localized information, so they captured the attention of many readers. The Newhouses recognized this and executed a local monopoly strategy. Newhouse newspapers were often the only paper in town—the only way consumers could read local news. The papers, which had the attention of an entire community, were a microphone with which to talk to it. Companies paid a premium for the right to use the microphone to speak to the community, which drove highly profitable advertising revenue. Once a local monopoly was established, it was hard to compete with. It would generate predictable revenues and cash flows for years and wouldn’t require much reinvestment from the Newhouse family. The result was an annual stream of cash that the family invested into other businesses.

The book was first published in 1994 so it doesn’t capture how the internet disrupted their local monopoly strategy, but I’d imagine it negatively impacted their newspapers as it did the rest of the industry. The internet made information more readily available and made it easier for companies to reach consumers in a specific community. Meta (Facebook), Alphabet (Google), and others are now enormous companies and still growing quickly. Much of their revenue comes from advertising, which went to newspapers before the internet era.

The Newhouse family still has an empire, but I’m pretty sure that newspapers are no longer its backbone. I’m curious about how the family adjusted their strategy to respond to the internet’s impact and about what business is now the empire’s backbone. The Newhouse family’s companies aren’t publicly traded so information isn’t available via the SEC, but I’ll do some research and see what I can find.

For my personal project, I’ve been reading biographies about entrepreneurs and documenting and sharing what I learned from each book. My mission for this project is to create a library of wisdom from notable entrepreneurs that I and other entrepreneurs can use.

I’ve been playing with ways to create this library. Each time I’ve finished a biography, I’ve created a digest—my interpretation of the book and its most essential elements. The digests contain ideas and experiences from the entrepreneurs’ journeys. I use them when I’m trying to solve my own entrepreneurial problems. That usually involves reviewing a digest to find relevant experiences of an entrepreneur. It’s better than rereading an entire book, which saves me time. But it’s still not great because leveraging multiple digests simultaneously means reading multiple digests again. For the sake of time, I usually have to pick the one or two that I think are relevant.

Creating a library of digests or entrepreneurial wisdom is helpful but not enough. It’s still too hard to use all that wisdom. I’ve been exploring ways to solve this. I’ve been learning about retrieval augmented generation (RAG). I’m still new to RAG, but my elementary understanding is that it’s a way to improve AI responses. RAG allows you to provide a knowledge base to AI to complement its large-language model (LLM). The result is more-accurate responses—or it’s supposed to be, at least.

If I can complement an LLM with multiple digests and ask AI for suggestions on solving my problem, I can leverage the wisdom of all the entrepreneurs I’ve read about to solve a single problem. Or at least that’s what I hope.

I’ve encountered an issue, though. Implementing RAG appears to require some technical abilities, but I’m not technical. I need to test my thesis and have been looking for ways to use RAG. This week, a friend told me about the latest updates to Google’s NotebookLM. NotebookLM is a productized way for people to leverage proprietary data and AI easily. It’s basically RAG made easy for nontechnical people like me. I’m oversimplifying because it does lots of other stuff, but you get the point. NotebookLM is my best option right now, so I’ll test my thesis using it until I can get some technical help.

NotebookLM has come a long way in the few months since I first heard about it. I’m excited to play with the latest version to test my thesis.