Blog

Thoughts on knowledge, tools, and AI.

Writing about personal knowledge management, building in public, and the ideas behind Harbor.

  1. knowledge-basebuilding-in-public

    Why I Built Harbor: The Case for a Private Knowledge Base

    The tools we use to manage knowledge are broken. Notes live in one app, people in another, tasks somewhere else — and nothing talks to your AI. Here's why I spent a year building something different.

  2. ai-trustaudit-trail

    The Case Against Trusting Your AI Blindly

    An AI agent deleted a company's entire database in nine seconds. It even apologized. The problem isn't AI malice — it's AI fluency.

  3. knowledge-managementnote-taking

    The half-second before you write it down

    Your knowledge base is not a record of what happened. It's a record of what cleared your threshold for significance — and the threshold is invisible to every tool ever built.

  4. securitymulti-tenancy

    The Workspace as a Security Boundary

    Access controls bolted onto shared databases are fragile. On why structural isolation — one file per user — is a more honest security model.

  5. knowledge-basenote-taking

    What Darwin Was Afraid to Forget

    Your knowledge base is full of things you agreed with. Darwin noticed this problem in 1876 and built a specific practice to fix it.

  6. knowledge-managementmemory

    You're Writing to a Stranger

    Notes are addressed to your future self. But research shows we experience future-us like strangers — and we write accordingly, to our cost.

  7. knowledge-baseai-agents

    The Quiet Power of Knowing What Your AI Doesn't Know

    Giving an AI access to everything turns out to be worse than giving it access to the right things. On why scoping is a performance decision, not just a security one.

  8. knowledge-managementmemory

    The Knowledge You Can't Write Down

    Notes capture what you thought. They rarely capture why — and almost never the judgment that decided it.

  9. ai-memorypersonal-knowledge

    Your AI Doesn't Know You — And That's the Problem

    AI assistants are more capable than ever, but every conversation still starts from zero — and that gap is bigger than most people realize.

  10. ai-agentsconfiguration

    The instruction you didn't write

    All system prompts are incomplete contracts. The question isn't whether there are gaps — it's whether you can see what fills them.

  11. personal-knowledgeai-memory

    The Knowledge AI Can't Generate

    As AI floods the web with synthetic content, public information is losing value. The knowledge that remains scarce is the kind that actually happened to you.

  12. knowledge-managementdesign

    The Blank Canvas Is a Trap

    Unlimited flexibility sounds like a feature. In knowledge tools, it usually isn't.

  13. ai-agentstransparency

    AI Agents Are Configuration, Not Magic

    Peel back the hype and an AI agent is a JSON record: a system prompt, a list of tools, a scope of what it can see. That's it.

  14. aitrust

    An Audit Trail Isn't Compliance. It's Trust.

    Most AI tools write to your data silently. What changes, psychologically and practically, when every AI edit is visible and reversible.

  15. data-ownershipportability

    The New Meaning of Exporting Your Data

    Export to CSV is theater. Real data portability means something you can restore from, self-host, and read in twenty years.

  16. engineeringsync

    Real-Time Sync Is Harder Than It Looks

    Behind every 'saved' indicator is a small distributed systems problem. What's actually happening when you type in a browser tab.

  17. ai-memoryknowledge-management

    The Reader in the Room

    When you connect an AI to your notes, you get a reader. And readers change what you write — whether you notice it or not.

  18. ai-memoryknowledge-base

    The Transcript Problem

    AI chat logs look like memory but work like conversation. The difference determines whether you can find anything later.

  19. ai-toolsthinking

    Tools That Think With You, Not For You

    The difference between AI that extends your thinking and AI that replaces it. On friction, brain connectivity, and why some slowdowns are features.

  20. knowledge-managementnote-taking

    Capturing Everything, Understanding Nothing

    More notes, fewer memories. The collector's fallacy is real — and AI tools might be about to make it much worse.

  21. knowledge-managementmarkdown

    The Document and the Database Were Always the Same Thing

    Plain text and structured data have been treated as opposites for fifty years. That's an engineering artifact, not a law of nature.

  22. knowledge-managementlearning

    The Capture Fallacy

    Saving information feels like knowing it. Research says otherwise — and that gap explains why most knowledge systems quietly fail.

  23. ai-memoryknowledge-base

    What the Summary Loses

    When AI summarizes your notes, it deletes the anomalies first, and the anomalies are usually where the knowledge lives.

  24. marginaliaknowledge-management

    The conversation in the margins

    Marginalia is the oldest knowledge management practice that actually works. Something about why gets lost in translation to digital tools.

Blog: Harbor | Harbor