Obsidian + Claude Is Game Over for Heptabase
A content creator tests migrating from Heptabase to Obsidian using Claude AI with the new 1 million token context window, demonstrating how Claude can set up PKM systems, create interconnected notes, and automate knowledge management tasks directly within local markdown files. The video explores the potential of AI-powered personal knowledge management while showcasing the 'My Life' organizational framework and discussing tool-agnostic systems.
Summary
The creator begins by explaining their journey through various PKM tools (Rome Research, Notion, Obsidian, Heptabase) and how Claude's capabilities, particularly Claude Opus 4.6 with its 1 million token window, have changed the landscape of knowledge management. They emphasize that while tools change, systems should remain constant, and that using only 20-30% of a tool's capacity while leveraging its superpowers is the optimal approach.
The core demonstration involves live-testing three key scenarios: (1) migrating a Heptabase backup into Obsidian while preserving markdown formatting, backlinking, and even whiteboard representations as canvas files, (2) setting up a new Obsidian vault based on the 'My Life' concept from their I Core course (which structures knowledge into Current Projects, Key Elements, and Topics), and (3) creating interconnected databases and canvas visualizations within Obsidian using Claude's understanding of the tool's structure.
Throughout the stream, the creator discusses their broader tool stack, which includes ClickUp for project management, Readwise/Reader for content curation, Gmail for email, MeetGeek for meeting transcription, Google Workspace for business, and Dropbox for cloud storage. They explain the importance of decoupling the 'outer world' (where content is saved) from the 'inner world' (the PKM system), using inboxes to process information before it enters the deep thinking system. They highlight how MCP (Model Context Protocol) connections allow Claude to understand context from multiple tools simultaneously, making semantic searches and automated workflows possible without traditional automation breakdowns.
Key demonstrations show Claude using local file access to create journal entries with AI-discovered insights, establish bidirectional relationships between notes, generate canvas visualizations of concept relationships, and potentially sync tasks between Obsidian and to-do applications. The creator encounters several technical challenges (slow performance, disconnected MCPs, stuck processes) but uses these moments to explain concepts like running multiple agents in parallel, the value of planning mode in Claude, and the importance of having proper documentation for AI to reference.
The creator concludes by emphasizing the paradigm shift: instead of building custom tools, users can now leverage AI's semantic understanding working directly with local markdown files in tools like Obsidian, eliminating the need for external servers while maintaining privacy and local control. They promise a follow-up video with deeper implementation details.
Key Insights
- Claude's 1 million token context window in Opus 4.6 enables AI to read and process entire PKM systems at once, making it possible to understand complex relationships and interconnections within knowledge bases that were previously too large for semantic processing.
- The creator argues that using only 20-30% of tool capacity while leveraging their superpowers, combined with proper system design, is more effective than trying to do everything within a single tool, regardless of how feature-rich it is.
- Decoupling the 'outer world' (where information is collected and stored) from the 'inner world' (the PKM system for deep thinking) prevents noise in the core thinking system and ensures only signal remains, with tools like Readwise handling the outer world and separate tools handling the action layer.
- AI with MCP connections can now understand semantic meaning across multiple disconnected tools simultaneously, making traditional automation rules unnecessary because the AI can interpret intent even with varied wording and apply context from ClickUp, to-do lists, and PKM systems in a single request.
- Markdown files in local folders accessible to Claude represent a fundamental paradigm shift—instead of building custom integrations or APIs, users can leverage AI's native ability to read and manipulate markdown to automate knowledge management while maintaining complete privacy and local control.
Topics
Transcript
[0:02] All right, here we are. And I cannot wait to dive into this today, guys, because this might be the moment where I'm going back to Obsidian. Who knows? Everything started with Rome research and notion and very soon and very soon Obsidian came out and that was it. Okay, there was backlinking born and all these things that we leverage very naturally in many and that we [0:34] expect in most of the PKM tools out there. And uh the thing is people always were talking about privacy and having their files local and so on but it was very restrictive access accessing these files across different platforms. Yes, we had the obsidian sync, but then when it…
Full transcript available for MurmurCast members
Sign Up to AccessMore from ICOR with Tom | AI Productivity
I Made Opus and Fable Grade Each Other. Opus Admitted It Lost.
The creator demonstrates real-world AI applications in their productivity business using Claude Fable and Opus, showing how Fable excels at comprehensive analysis of complex systems while comparing its performance to Opus on specific tasks. The key finding is that Fable's holistic understanding justifies its higher cost for business-critical audits and optimizations.
Our AI invented the numbers. We caught it.
The hosts discuss the importance of double-checking AI outputs and validating information, emphasizing that this is not a new problem but rather a longstanding practice required with any information source. They explore how AI tools can be customized to individual workflows and contexts, arguing that understanding foundational methodology matters more than specific tools.
An AI Just Handed Me a Fake $67B Statistic
The speaker shares how an AI confidently generated a completely fabricated $67.4 billion statistic while researching AI hallucinations, then explains his systematic approach to combating AI misinformation: using dual independent search engines to cross-check answers and identify where hallucinations occur.
AI Is Quietly Taking Your Memory
AI companies are racing to own users' context and memory through integrated tools like Claude in Slack and ChatGPT's auto-memory features, creating a lock-in trap. The speaker argues users should own their context (expensive, irreplaceable) in local plain text files while renting the AI model (cheap, replaceable), and demonstrates his personal knowledge assistant system built on markdown files that work with any AI model.
My AI Team Now Has an Interface. All 12 Agents. Free.
Tom, creator of the iCola methodology, releases My PKA version 3 with a free interface (My PKA Cockpit) that provides 12 pre-built AI agents for productivity, note-taking, health tracking, document management, and task planning—all running locally in a folder structure accessible via Claude and a web interface.