Heptabase killed NotebookLM and Mem (Best AI Note-Taking App 2026)
The speaker compares Heptabase's newly released AI features with NotebookLM and Mem, arguing that Heptabase's AI is superior for personal knowledge management because it leverages structured organization rather than relying entirely on AI to find connections. The key difference is that Heptabase integrates AI into an existing knowledge architecture, whereas NotebookLM requires users to upload sources without inherent structure.
Summary
The speaker begins by explaining that Heptabase removed its initial chatbot feature years ago to implement AI properly and meaningfully rather than as a superficial addition. According to community data from the iCor tool finder, Heptabase ranks at the top for user adoption in personal knowledge management, with 13% of their community using NotebookLM as a comparison tool.
The core argument centers on how structure affects AI performance. The speaker demonstrates this using a personal knowledge base about iCor methodology containing 127 sources (YouTube videos, articles, podcasts, and a published book). When asking questions like "What is the capturing beast?" in Heptabase, the AI provides answers with proper source attribution and clickable references that show exact sections where information originated.
The speaker identifies a fundamental difference in approach: Heptabase requires users to build organized whiteboards with interconnected cards and clear hierarchical structure before leveraging AI, whereas NotebookLM accepts sources in a flat list format. The speaker argues this matters significantly because in NotebookLM, users rely 100% on AI to determine relevance, while in Heptabase, the pre-existing structure guides AI queries toward more accurate results.
A practical comparison test is conducted using two YouTube videos about iCor in both platforms. Both tools provide comprehensive answers when asked "What is iCor?" However, Heptabase's responses include timestamp-specific video links that jump to exact moments in the source material, while NotebookLM shows transcript sections requiring manual navigation. Additionally, Heptabase allows users to create new cards directly from AI responses, which automatically maintain connections to source materials and can be reused across multiple whiteboards in different contexts.
The speaker acknowledges NotebookLM's strengths with vast knowledge bases but states Heptabase "killed" NotebookLM for their personal use case. They also note that Heptabase previously didn't support Google Docs links (a NotebookLM advantage), though they hope this improves. The speaker briefly dismisses Mem as inferior to Heptabase for knowledge management, noting that Mem's video embedding requires extensions and lacks smooth integration.
The concluding argument frames the key distinction: Heptabase transforms AI into a "thinking partner" and "brainstorming partner" within an already-organized knowledge system, rather than making AI the primary driver of organization. The speaker invites viewers to discuss specific NotebookLM features they prefer and offers to conduct further testing with larger knowledge bases.
Key Insights
- Heptabase deliberately delayed implementing AI features to ensure they would be meaningful rather than superficial, based on founder Alan's philosophy that AI should be useful for users rather than just adding a basic chatbot
- The speaker argues that Heptabase's AI superiority stems from pre-existing structured organization—users who build interconnected whiteboards before querying AI receive better results than those who rely on AI to find connections in unstructured lists
- In Heptabase, clicking on AI-provided source references jumps to exact timestamps within videos, whereas NotebookLM shows transcript sections requiring manual navigation to find context
- NotebookLM excels with vast knowledge bases requiring deep understanding across many sources, but Heptabase killed NotebookLM for the speaker's use case of lifelong personal knowledge capturing and retrieval
- Heptabase allows AI-generated summaries to be converted into reusable cards that maintain source connections and can be applied across multiple whiteboards in different contexts, whereas NotebookLM confines information within single notebooks
Topics
Transcript
[0:00] Hey everybody. Today we will talk about Heepbase and how it really catched up with notebook LM and even MEM. And I ditched me a long time ago for Hepabase, but I've been also using Notebook LM for a long time. Make deep research on the knowledge base that I provided. And Heepbase was missing AI features for a long time. They had an AI chatbot in the beginning and then they removed it. And if you follow us and you watch the podcast episode where we interviewed Alan, the founder of Heptterase years ago, they said if they implement AI, [0:32] they want to make it properly and it should be useful for the people and not just…
Full transcript available for MurmurCast members
Sign Up to AccessMore from ICOR with Tom | AI Productivity
Claude sucks. And here is why.
This video explores why Claude and other AI models produce inconsistent results, examining how model selection (Haiku vs Sonnet vs Opus vs Claude 3.5 Fable), effort levels, prompt clarity, and system context dramatically affect output quality. The speaker demonstrates these differences through a Venn diagram creation task and advocates for using organized folder systems with documented SOPs and code-based solutions to ensure consistent AI performance.
Claude can now prompt its own SESSIONS. I built the boss.
Claude now has the ability to interact with multiple sessions, allowing a streamlined orchestration of tasks through an agent called Larry. This integration can simplify task management and enhance productivity by enabling a single main session to oversee various sub-sessions or tasks.
I switched off Claude's memory. My folder does it better.
Tom explains why he disabled Claude's native memory feature and instead built a local folder-based system (myPKM) to maintain persistent context with AI agents. He shares an audit revealing systemic issues like contradictory documentation, forgotten rules, and parallel session conflicts, then outlines how to systematically fix these problems using the ICOR productivity methodology.
I do not download invoices anymore. My folder and Claude do it.
The speaker demonstrates how he uses AI (Claude) integrated with his local folder system to automate invoice collection for his accountant, downloading 36 invoices in 5 minutes with minimal manual intervention. He explains his AI team methodology, work streams, and how persistent local knowledge bases enable AI to handle recurring tasks independently without constant re-explanation.
Claude finished the work. Then I lost it in the chat.
A creator demonstrates a comprehensive local-folder-based system for managing AI-generated deliverables asynchronously, featuring an interactive browser interface that tracks decisions, changes, and conversations without relying on external databases or web applications. The system enables collaborative editing, visual annotations, and automatic context preservation across multiple parallel AI projects.