Blog

Where does your knowledge live?

Take a moment and think about where your knowledge actually lives. Your notes are probably in one app, your documents in another, your saved articles in a third. Your highlights from books are in Kindle. Your bookmarks are in Chrome or Safari. Your research lives in Zotero or Mendeley. Your screenshots are in your camera roll. Your voice memos are in whatever your phone's default recorder is. Your work documents are in Google Drive or Dropbox or SharePoint, possibly all three. Your Slack messages contain years of decisions, context, and institutional memory that you'd struggle to reconstruct if they disappeared.

Each of these systems holds a piece of what you know. None of them hold the whole picture, and none of them were designed to work together. Your knowledge is distributed across a dozen landlords, each with their own terms, their own export limitations, their own business model, and their own views about what you're entitled to take with you if you leave.

Most people have never sat down and thought about this, in the same way most people don't think about where their pension is invested until something goes wrong. It's fine until it isn't, and by the time it isn't, the options are limited.

The landlord problem

Every app you store knowledge in is a landlord, and the terms of the tenancy are set entirely by the landlord. This sounds dramatic until you remember what's actually happened over the past fifteen years.

Google Reader was the centre of many people's information workflow. Google shut it down in 2013 with a few months' notice, and the ecosystem of reading and sharing that had built up around it simply vanished. Evernote was the default knowledge management tool for years, and its slow decline left millions of users with notes trapped in a proprietary format that no other tool could cleanly import. Photobucket decided to hold users' images hostage behind a $400/year paywall, breaking embedded images across the entire internet. Notion changed its pricing. Roam Research went quiet. Every tool you depend on is one business decision away from becoming a problem.

The common advice is "you can always export your data," and technically this is often true. GDPR's Article 20 established data portability as a right, at least in Europe. But anyone who has actually tried to export from a major platform knows that the export is usually a dump of files in formats that no other tool can easily ingest. You get your data in the same way you'd get your belongings back from a flat you've been evicted from: technically complete, practically useless without significant effort to reorganise.

The deeper issue is that even when your data is exportable, the structure and connections you've built up over years typically aren't. Your folder hierarchy, your tags, your links between notes, the search index that lets you find things by meaning rather than filename: all of that is platform-specific. Moving to a new tool means starting the organisational work from scratch, which means most people don't move, which means the landlord's leverage only grows over time.

The attention tax

There's a subtler cost to having your knowledge scattered across platforms, beyond the lock-in risk. Every time you need to find something, your brain runs a small routing decision: which app is this in? Did I save it in Notion or Google Docs? Was it a bookmark or a web clip? Did I email it to myself or save it to Pocket?

These micro-decisions add up. Research suggests knowledge workers spend 45 minutes per day just searching for information they need. That's partly a search problem, but it's substantially a fragmentation problem: the information exists, you just don't know which of your six systems it's in. And each failed search attempt in the wrong app costs time, attention, and a small amount of the cognitive energy you need for actual work.

The fragmentation also prevents connections. An article you saved in Pocket, a note you wrote in Apple Notes, and an email thread in Gmail might all relate to the same project, but they'll never surface together because the tools don't share context. The insight that would come from seeing them side by side, the connection your brain would make if they were in the same field of vision, simply never happens.

Where your data should live

The question of where your knowledge should live has a surprisingly simple answer, even if the implementation takes some thought: it should live in a place you control, in a format you can move, accessible to whatever tools you choose to use today and whatever tools you might choose tomorrow.

This means a few things in practice.

Storage you own. Your files and documents should live in storage that belongs to you rather than to the app you're using to view them. Bring-your-own storage (connecting your own Amazon S3 bucket or Cloudflare R2) means the data physically sits in infrastructure you control. The application layer reads from and writes to your storage, but if you switch applications, the data stays exactly where it is. You're choosing a lens, not a landlord.

Portability by design. The system should use open formats and open protocols so that moving to a different tool doesn't require a painful migration. MCP (Model Context Protocol) is one example: it lets any AI agent read from your library on your terms, without the data being absorbed into the agent's own systems. The library persists regardless of which model or which tool you're using, because the library belongs to you and the tools are interchangeable.

Model choice. If your knowledge system uses AI (and increasingly they all do), you should be able to choose which models it uses, including open source models that run without sending your data to a third-party API. The model should serve you without absorbing your data as a side effect of the service.

One searchable library. Rather than accepting fragmentation as the default, your knowledge should flow into one place where it's searchable by meaning regardless of where it originated. Your Google Drive, your Dropbox, your email, your reading highlights, your native notes, your web clips, your voice memos: all findable from one search bar, all connected, all yours.

The knowledge audit

If you've never done this exercise, it's worth ten minutes.

Open your phone and your computer. List every app and service that holds something you'd consider knowledge: notes, documents, bookmarks, saved articles, highlights, research, screenshots, voice recordings, emails you've kept for reference. For each one, ask three questions: could I export everything from this right now? If this service shut down next month, what would I lose? And does any other tool I use know that this information exists?

Most people find the answers uncomfortable. The export is theoretical rather than practical, the potential loss is larger than expected, and the tools are almost completely siloed from each other. Years of accumulated thinking, reading, and research, spread across a dozen systems that each hold a fragment of the picture while none of them hold the whole thing.

The good news is that consolidating doesn't require a dramatic overhaul. Start by connecting your existing sources to one searchable library rather than migrating everything manually. The data can stay where it is while becoming findable from one place. Over time, the library becomes the centre of gravity, and the individual apps become inputs rather than destinations.

The underlying principle is the same one that applies to any valuable asset: know where it is, make sure you can access it on your terms, and don't leave it entirely in someone else's hands. Your knowledge has taken years to accumulate. It deserves at least as much care as you'd give your financial records.

Frequently asked questions

Isn't this just the same as backing up my files? Backups protect against data loss. What we're describing protects against something different: data fragmentation and platform dependency. A backup of your Notion workspace is still a Notion backup, useful only if you're still using Notion. A knowledge library that lives in storage you control, in open formats, with portability built in, protects against the platform changing, shutting down, or deciding to compete with you.

Do I really need to worry about platform lock-in? History suggests yes. Google Reader, Evernote, Photobucket, Sunrise Calendar, Wunderlist, Inbox by Gmail: all widely used, all gone or fundamentally changed. The pattern is reliable enough that building your knowledge system on the assumption that your current tools will exist unchanged in five years is a bet worth hedging.

What about Google Drive or Dropbox as the central store? They're fine as storage backends, less reliable as the organising layer. Google Drive doesn't understand the meaning of your documents, can't connect a PDF you saved six months ago to a note you wrote yesterday, and doesn't let you search by concept. A knowledge library with semantic search adds the intelligence layer on top of whatever storage you already use.

How does bring-your-own-storage work? You connect your own cloud storage (Amazon S3, Cloudflare R2, or similar) to the application. Your files live in your storage account, under your control, with your encryption keys. The application reads from and writes to that storage, but doesn't hold the data itself. If you stop using the application, the data remains in your storage exactly as it was.