From Generic Chatbots to Personal Knowledge Bases
AI personalization features are systems that let an assistant remember your data, preferences, and projects over time so it can deliver answers, drafts, and actions that match your ongoing work instead of treating every session as a blank slate. Google’s NotebookLM and Perplexity’s Brain memory system both show this shift away from one-off prompts toward persistent, custom knowledge base AI. Rather than pulling only from the public web or a single upload, these tools are starting to organize long-term context tied to how each person reads, writes, and researches. That changes the value of an AI assistant: it becomes a companion to your workflow, not just a search box with better language skills. The question now is how far this memory layer will extend—and how clearly users can see and control what is stored.
NotebookLM’s Personal Intelligence and AI Note Editing
Google is preparing two upgrades that turn NotebookLM into more than a neutral reader of your documents. The first, called Personal Intelligence, is a NotebookLM memory system that pulls context from past conversations inside the app, keeps it, and reuses it later. It is scoped tightly: according to TestingCatalog, it “appears to learn only from activity inside NotebookLM rather than reaching across Gmail, Docs, or other Google surfaces,” a guardrail aimed at researchers and teams with strong privacy expectations. Controls to switch it off and inspect stored data are also surfacing. The second feature, AI Editing for notes, links the chat and noteboard more closely. You can highlight text in a note, send it into chat as context, and ask NotebookLM to rewrite or refine it, solving a long-standing limitation where saved responses could not be edited once created.
Inside Perplexity Brain: A Transparent Memory Layer
Perplexity is building Brain as a shared memory substrate that sits beneath multiple products, including Perplexity Search and the Perplexity Computer agent. Rather than hiding what it knows, the Perplexity Brain tool is designed to expose its knowledge in three ways: topics grouped into categories, the detailed context behind each topic, and a 3D map of connections you can hover over and explore. This layout mirrors “second brain” apps like Obsidian, where information is clustered instead of shown as a flat list. Because Brain retrieves context by topic only when a task needs it, each request can draw on a smaller, more relevant slice of memory, which should help both speed and answer quality. The feature is still hidden, but recent polishing suggests Perplexity is getting ready for broader use, turning Brain into a shared context pool across its search, agents, and browsing.
Why Persistent Memory Is the Next Competitive Battleground
Both NotebookLM and Perplexity Brain move AI tools away from stateless, generic replies toward assistants that remember and adapt. In practice, this means your AI agent can recall earlier research threads, writing styles, or technical setups without forcing you to repeat long instructions at the start of every session. That reduces friction and makes it easier to treat the AI as part of your daily workflow, not a one-off novelty. It also marks a competitive shift: companies are racing to turn memory into a core platform layer rather than an optional feature. Perplexity is wiring Brain across search, its Computer agent, and possibly its Comet browser, while Google is folding NotebookLM into a broader Gemini ecosystem. The long-term winners are likely to be the tools that combine strong memory with clear controls, transparency, and tight integration into the documents and tasks people care about.
