You spend ten minutes explaining who you are and exactly how you want your information formatted, only to have to do it all over again the next time you open a new chat. It is tedious. But if a recent discovery is anything to go by, Google is about to fix this frustration for its research tool, NotebookLM.
In early February 2026, TestingCatalog spotted something interesting hidden in Google’s code. They found a feature called Personal Intelligence being tested inside NotebookLM.
This isn’t an official announcement yet, but the leaked settings suggest a shift is on the way.
To understand why this matters, you have to look at what NotebookLM is right now. Unlike a standard chatbot that searches the whole web, NotebookLM is a grounded AI.
You upload your own PDF files, notes, or transcripts, and it answers questions using only that material. It is excellent for students and researchers because it doesn’t usually make things up.
But currently, it is a bit of a blank slate. Every time you start a new notebook, the AI forgets your preferences. You are back to square one.
The Personal Intelligence update aims to give the AI a long-term memory. According to the findings, the system will learn from your past chats to understand your goals. It creates a persona for you.
For example, if you are a biology student who prefers simple summaries, the AI will remember that preference across different sessions. You won’t have to keep typing “explain this simply” every time.
The leaks show you can set these preferences globally for the whole app, or tweak them for specific notebooks. You could have one persona for serious legal research and another for a creative writing project.
Here is where the speculation gets interesting. Google has already rolled out a version of Personal Intelligence for its main Gemini chatbot, which has access to your Gmail and Drive to find information.
However, the NotebookLM version seems different. It appears to be bounded. This means it only remembers what you do inside NotebookLM and doesn’t go through your emails or other Google apps.
For professionals and businesses who worry about data privacy, this is a relief, as the AI only knows what you explicitly give it to work on.
Personal Intelligence solves the pain point of repetition. If the feature ships as tested, you could define your style once, like “always cite sources” or “write in conversational style”, and the AI will apply that standard everywhere.
While we don’t have a release date yet, this test shows that the future of software isn’t just about being smarter; it is about being more adaptive to the humans using it.