Have you heard of NotebookLM? It’s an AI-powered tool from Google that lets you upload a bunch of documents into a virtual notebook, and ask it questions. I use it a lot in my work. It’s super helpful for integrating information across multiple documents, PDFs, and slide decks.
Instead of sifting through different sources when I need some info, I dump all the documents into NotebookLM, and ask it questions. So I could ask a question like “What are the asset requirements for setting up a Performance Max campaign?” and NotebookLM will pull up the info for me. And since the response is based on my own data (and NotebookLM links to the exact source in each response), there’s a much lower risk of it hallucinating a wrong answer.
Here’s a secret: Sometimes, when a client asks me a really specific question about a product, I open NotebookLM and pull out the answer in real time. Then I’ll casually give the answer as if I’m really knowledgable about esoteric topics like Conversion Lift Studies.
It’s gotten me thinking about how AI can give you superpowers – but only if you give it the right source data.
AI Works Best With Your Own Data
Chatbots like ChatGPT and Gemini are fine for general use cases, but sometimes the best answers come from your own sources. Think about the notes you captured from that one meeting, the health screening report you received 3 years ago, or that email you received from that unknown Singaporean blogger who used to write about personal finance but now can’t seem to find a writing focus. Information like this doesn’t exist in ChatGPT, but is often way more personalized and valuable to you.
In the past, search used to be the only way to retrieve this info. We’d search our Gmail and our Documents folder, trying different combinations of keywords (Did I save that file as “presentation_final”, or “final_prez” or “presentation_v4_final_final_2”?)
But with AI, we can now directly extract the information that’s most relevant for us. Like NotebookLM, platforms have already launched AI features to help us engage with our own source data:
- Evernote released an AI-powered Search function where you can ask it questions like “What are the to-dos from yesterday’s meeting?” or “What was that article about incrementality and giving out flyers?” (I used this particular prompt for my previous post!)
- The Ask Photos feature on Google Photos allows you to ask it questions like “Show me the best photos from each beach I’ve visited” or “Where did we have that delicious qaesadilla last year?“
- Gen AI platforms are focused on increasing the number of tokens you can upload to give them more context. For example, Gemini 1.5 Pro will soon have a 2-million token context window, which means you can upload a 2-hour long video or a 1M-word transcript, and ask it questions.
In a world where everyone is using AI, the one thing that will give you superpowers is the source data you feed it.
How To Tap Into This Superpower
However, most of us don’t have a personal repository of high-quality source data. Think about all the great books, insightful articles, or engaging videos that you’ve read but forgotten. Think about all the in-depth conversations you’ve had years ago, but only vaguely remember today. We encounter amazing experiences and learnings every day, but we’re not great at capturing them.
How powerful would it be if you could retain this information, and use AI to sift through them to find the most relevant, valuable insights?
We could start with simply capturing more about what’s valuable or helpful to us.
- Take copious amounts of notes and store them digitally in a platform like Notion, Evernote or Apple Notes
- Use a tool like Readwise to capture highlights from ebooks or articles
- Save as many emails and photos as possible.
Yes, this data probably sits across multiple sources today: Your Google Drive, your note-taking app, or your email account. And you might have to fork out a few extra bucks to store all of this data.
But I’m willing to bet that the tech will soon allow us to aggregate multiple sources of data. Imagine how cool it would be to ask AI to extract your health reports from your Google Drive, cross-reference them to your meal pictures on Photos, and see if your diet contributed to your higher cholesterol levels over time.
As AI continues to advance, the role of personal source data will become even more important. But building up a personal repository of data takes time, so it might be a good idea to start now.
