What is a personal knowledge base? A plain guide to building one that stays useful

A personal knowledge base is a private place to keep and find what you know

A personal knowledge base (PKB) is a private, searchable store of things you want to remember: notes, links, recommendations, and facts you can find later. It is the one place you go when you know you saw something but can't recall where.

People keep one because their useful stuff is scattered. A restaurant a friend swore by, a link you meant to read, an intro you promised, a hotel someone booked. It all lives in different apps.

The hard part is not storing all of it. The hard part is keeping it current, so you still trust it a month from now.

Why people want one: the cost of information you can't find

Think about the last recommendation you lost. A friend texted you the name of a great restaurant, and now it sits buried a thousand messages back.

You know it exists. You just can't put your hands on it when you need it.

This adds up, and it has been measured. A 2012 McKinsey report found workers spend 28 percent of the workweek on email and nearly 20% of the workweek on internal information. That was before most of today's apps existed.

The problem grew as our tools multiplied. Workers spend an average of 3.6 hours every day searching for information at work, according to a 2022 Coveo survey the company sponsored.

The same survey found 60% of employees have to search within four or more data sources every day. Every one of those sources is another place to check.

A Harvard Business Review study tied the average user's switching apps 1,200 times a day to 9% of their annual time at work.

Memory alone won't hold it: why you need something outside your head

You might think you'll just remember the important things. Most of us won't, and the science on this is old and steady.

In 2015, researchers Murre and Dros published a 2015 replication study in the journal PLOS ONE. They concluded that the Ebbinghaus forgetting curve has indeed been replicated.

Here is the honest caveat from that work. Ebbinghaus tested himself on nonsense syllables, a worst case, so meaningful things fade more slowly than his numbers suggest.

Even with that caveat, the pattern holds. Memory needs help, and a PKB is that help: the store that lives outside your head.

Personal knowledge base vs personal knowledge management vs a notes folder

People mix up three terms. Sorting them out makes everything else easier.

A PKB is the store itself. Personal knowledge management (PKM) is the practice of running that store by capturing, organizing, and reviewing what's in it.

A plain notes folder is a PKB with a weak spot. You can put things in, but you can't easily ask it questions or see how notes connect.

The common methods: second brain, PARA, and Zettelkasten

Once people commit to a PKB, they usually reach for a named method. A few come up in almost every guide, so here is what each one means.

These methods work on paper. They share one catch: every one of them is a filing system you run by hand.

• Second brain: A system for storing ideas outside your head so you can act on them later. It fights the fear of forgetting.
• PARA: A way to sort everything into Projects, Areas, Resources, and Archives. It answers the "where does this go?" question.
• Zettelkasten: A method of small linked notes that reference each other. It fixes the problem of ideas sitting in isolation.
• Maps of Content: A single note that links out to related notes, like a table of contents you build by hand.

The tools most people use, and where they stall

Most people pick a tool before a method. Here are the ones that come up most, and what each does well.

Each of these is good at its job. They share the same dependency: you have to enter the notes and keep them up to date.

Privacy usually shows up as a line item in these comparisons, not the headline. Obsidian keeps files local by default, while most of the others sync to a company's servers.

• Notion: Strong for structured databases, pages, and sharing with a team.
• Obsidian: Local-first plain text with fast linking between notes.
• Evernote: A long-running notebook for clipping and storing web pages.
• Roam: Outlining and networked notes for people who think in connections.
• Apple Notes: Simple, fast, and already on your Mac and phone.

The real problem: a knowledge base you have to maintain stops being true

Here is the part most guides skip. A knowledge base you maintain by hand goes stale the moment life gets busy.

A busy month happens, the entries stop, and the tool quietly becomes wrong. That is worse than having nothing, because you still trust it.

Manual upkeep has a second, hidden tax. Every time you stop work to file a note, you pay a switching cost.

The American Psychological Association notes that even brief mental blocks from shifting between tasks can cost up to 40% of productive time on complex work.

So the fix has to be automatic. The base should fill itself from things you already do, with no extra filing.

A different approach: a knowledge base built from your conversations

Look at where your useful information already lives. Most of it is sitting in your messages right now.

Someone texted you the restaurant. You promised in a thread to send a link. A friend shared an article worth reading, and you meant to open it later.

A tool can read that history on your Mac and turn it into something you can search. The base fills itself, because the raw material is already written down.

Sanctuary is one example of this approach. It is a Mac app that reads your own Messages history and builds a knowledge base you don't maintain out of it.

You can find what people sent you by asking a plain question. The links people text you turn into a read-it-later from messages, collected without you saving each one.

How retrieval works when you don't file anything

With a folder system, finding something means remembering where you filed it. You hunt through folders and tags until it turns up.

With plain-language retrieval, you ask a question instead. You type "who recommended that sushi place," and you get the answer along with the message it came from.

The same memory can resurface things, not only store them. It can surface the promises you made so the intro you owe a friend doesn't slip away.

Because it already knows your people, it doubles as a personal CRM you skip filling in. You never open a form to log a contact.

Privacy: why "local by default" matters for a knowledge base

A record of your conversations is sensitive. Where it lives matters as much as what it can do.

Local by default means the library stays on your Mac. There is no account, no sync, and no telemetry quietly sending your data somewhere else.

You don't have to take that on faith. Turn off your wifi and the app still answers, and you can check the privacy claims yourself.

Cloud AI stays optional. When you want a bigger model, it runs on a key you provide, and it is off until you turn it on.

Letting your AI assistant use it, on your terms

A knowledge base gets more useful when your AI assistant can ask it about your life. General assistants know the whole internet and nothing about you.

You can connect your knowledge to Claude with sharing you control. You decide what crosses: only what the app learned, or the full text of messages.

How to build a personal knowledge base that stays useful

You can build one with any tool. These steps matter more than the method you pick.

The more of this that happens automatically, the more your base stays true. A base you trust is a base you actually use.

I built Sanctuary because I kept failing at the manual version. My notes apps went stale every time work got busy, and I was tired of losing the same recommendation twice.

— Ravi

• Decide what to capture: Pick the few kinds of things you keep losing, like recommendations, links, and commitments.
• Make capture nearly effortless: The less work it takes to save something, the more likely your base stays complete.
• Keep one place to look: Scattered notes recreate the original problem, so consolidate into a single searchable home.
• Favor retrieval over organization: Spend your energy on asking good questions, not on building perfect folders and tags.
• Review only what resurfaces: Let the system bring back promises and dates on its own, instead of re-reading everything.