According to XDA-Developers, Google’s NotebookLM recently added a Custom mode feature that lets users program the AI to communicate in their specific writing style and thinking patterns. The author successfully created an AI version of themselves by configuring the tool to match their casual communication habits, including avoiding capital letters and using shorthand. They used this customized AI to explain technical topics like database creation in LibreOffice and even created a mental health course using their preferred bottom-up thinking style. The customization process involves clicking the settings icon in the chat panel, selecting Custom mode, and inputting up to 10,000 characters of instructions. The author found the results surprisingly effective across different topics, requiring only minor tweaks to perfect the AI’s mirroring of their communication patterns.
Why this actually matters
Here’s the thing – most AI customization features feel pretty superficial. You can change the tone a bit, maybe make it more formal or casual, but it still feels like you’re talking to a robot. NotebookLM’s approach is different because it actually lets you program how the AI thinks, not just how it talks. The author mentions programming their ADHD thinking patterns into the AI – short information bursts, looping back to previous points, mixing casual and proper terms. That’s way beyond just changing the vocabulary.
And honestly, this could be huge for people who struggle with traditional learning methods. When an AI explains complex topics using your own mental shortcuts and familiar language patterns, it removes that cognitive friction of translating “corporate speak” into how you actually think. The author mentions it helps avoid perfectionism spirals – if your AI research assistant can explain quantum physics using the same messy language you use with friends, suddenly that pressure to sound smart just evaporates.
The real-world applications
Think about how this could change professional training and education. Instead of forcing everyone to learn through the same corporate-approved communication style, you could have AIs that adapt to how different teams actually think and communicate. Marketing teams could get explanations in creative, big-picture language while engineering teams get technical, detail-oriented responses – all from the same underlying information.
The author tested this with some pretty diverse topics too – from technical database tutorials to mental health courses. And the customized AI handled both effectively because it was working within the author’s established thinking framework. That’s the real magic here – it’s not about creating a perfect AI, but creating an AI that works perfectly for you.
The future of personalized AI
What’s interesting is how simple the setup process seems. You’re not training a model from scratch or writing complex code – you’re just giving clear instructions about how you communicate. The author used bullet points and included some official terminology to guide the AI. Basically, you’re teaching the AI your personal “language” rather than learning its language.
I wonder how far this could go though. Could we eventually have AIs that not only match our communication style but actually anticipate our knowledge gaps based on how we typically process information? The author mentions this helps “keep you honest” about your own shortcomings – that’s pretty profound when you think about it. An AI that knows how you think might be better at spotting where your thinking goes off track.
This feels like a step toward AI tools that actually adapt to humans, rather than humans adapting to AI tools. And given how many people struggle with traditional educational and professional development methods, that could be genuinely transformative.
