Personal knowledge management in the AI era

Personal knowledge management is the practice of capturing, organizing, and retrieving what you learn so you can use it later. In the AI era, the practice gains a new payoff: when your notes are clean, owned, and stored as plain text, an AI assistant can read them and reason over them. Your knowledge base stops being just a place you search by hand and becomes context you can hand to a model. The core habits do not change. What changes is what a good note is now worth.

This post is about that shift. It is not a claim that AI replaces note-taking, or that you should pour everything into a chat box. It is a practical look at which personal knowledge management habits still matter, which ones matter more now, and why the file format under your notes turns out to be the deciding factor.

What is personal knowledge management?

Personal knowledge management is the set of habits and tools you use to turn scattered information into a body of knowledge you can act on. It usually covers four steps:

• Capture. Write things down: ideas, quotes, meeting notes, things you read.
• Organize. Give notes structure so you can find them: folders, links, tags, titles.
• Connect. Relate notes to each other so ideas compound instead of sitting alone.
• Retrieve. Get the right note back at the moment you need it.

People have run this loop on index cards, in notebooks, and in apps for decades. The "second brain," a term from Tiago Forte, is one popular framing: an external, trusted store for what your biological memory cannot hold. None of that is new, and none of it goes away because AI arrived.

How does AI change personal knowledge management?

The change is narrow but real. For most of PKM's history, you were the only reader of your notes. You wrote them for your future self, and retrieval meant you searching, scanning, and remembering. An AI assistant adds a second reader, one that can read your entire knowledge base in seconds, summarize across hundreds of notes, find connections you missed, and answer questions grounded in what you actually wrote.

That turns your notes into something they were not before: queryable context. Instead of just searching for a file, you can ask a question and get an answer drawn from your own material. Large language models are strong at this when they are given good source text. Point a model at a folder of clear, factual notes and it can draft, compare, and explain using your knowledge rather than its generic training. We go deeper on that mechanic in Markdown as AI context.

Here is the honest part. AI does not fix a messy knowledge base. If your notes are vague, contradictory, or trapped in a format the model cannot read, you get vague, contradictory answers back, and you spend your time correcting the model instead of using it. AI raises the ceiling on a good knowledge base and exposes the floor on a bad one.

What stays the same, and what changes?

Most of the discipline carries over. A short comparison:

PKM habit	Before AI	In the AI era
Capture	Write for your future self	Still write for your future self; a model reads it too
Organize	Folders and titles help you find notes	Clear structure also helps a model navigate them
Own your data	Nice to have; avoids lock-in	Essential; you can only feed a model files you control
Format	Any format your app supports	Plain text wins, because models read text natively
Retrieval	You search and scan	You search, and you ask questions across notes
Writing quality	Sloppy notes cost you later	Sloppy notes cost you twice: once for you, once for the AI

The pattern is clear. The fundamentals stay. Their stakes go up. A clear title was always polite to your future self; now it is also a signal a model uses. Owning your files was always wise; now it is the gate to using AI on your own material at all.

Why does plain text matter more now?

Because an AI model reads text. That sounds obvious, but it shapes how you should store knowledge.

A Markdown file is plain text with light, readable structure: # for headings, - for lists, **bold**, tables, links. There is no proprietary container around it. A model ingests it cleanly, headings and all, and the structure actually helps it understand what is what. Notes locked inside a database, a binary file, or a closed app are harder to extract and feed in. You end up exporting, converting, and losing formatting before the AI ever sees them.

Plain text also ages well. The format has been readable for fifty years and will outlive any single app. That durability mattered for PKM long before AI, and we made the full case in building a second brain in plain text. The AI era just adds a second reason: the most future-proof format is also the most AI-readable one. You are not choosing between longevity and usefulness. They point the same way.

There is a trade-off worth naming. Plain text gives you less out-of-the-box automation than a heavy database app. You do not get rich relational views or kanban boards for free. What you get instead is portability, durability, and a knowledge base any tool, AI included, can read without permission. For knowledge you intend to keep for years, that is the better deal.

What should you actually do?

Here is a setup that works whether or not you ever attach an AI assistant to it. None of it depends on a particular product.

1. Keep one folder of Markdown files. Notes, projects, and references as plain .md text you control. One source of truth, not five half-synced apps.
2. Write clearly, not cleverly. A good note states what it is in the first line. Clear notes serve your future self and any model you point at them.
3. Use light structure. Headings, short lists, and tables make a note scannable for you and parseable for a model. You do not need a rigid system.
4. Title and date your notes. A descriptive title and a date make a note easy to locate, by hand or by query. That is most of organizing, done up front.
5. Clean up AI-generated text before you keep it. If a model drafts notes for you, trim them to what is true and useful so you do not pollute your own base. More on that in clean up AI Markdown.
6. Own the files. Store them where you can read them without an account. That keeps your options open for whatever AI tools come next.

Notice what is missing from that list: any specific AI product, any subscription, any lock-in. The point is to build a knowledge base that is valuable on its own and ready for AI on top of that, rather than betting your knowledge on one vendor's features.

Does this mean dumping everything into a chat assistant?

No. A chat window has no memory of your knowledge base unless you give it one, and pasting notes one at a time does not scale. The durable approach is the reverse: keep your knowledge in files, and let tools read those files when you need them. The knowledge base is the asset. The AI is a reader that comes and goes.

This is why "own your files" has quietly become the most important PKM principle of the moment. You can switch AI tools freely when your notes live as plain text in a folder you control. You cannot if they are trapped in an app that may change its terms, its format, or its existence. The knowledge outlasts the tool, which was always the goal of a second brain.

The takeaway

Personal knowledge management in the AI era is the same craft with higher stakes. Capture, organize, connect, retrieve, and keep doing all of it. Write clearly. Use light structure. Above all, own your files and keep them as plain text, because that single choice is what makes your knowledge both durable and readable by the AI tools you will use this year and next.

Noteline is built on exactly this idea: every note is a plain Markdown file in a folder you pick, on your machine, with no account and no lock-in. That keeps your knowledge base yours, and ready for whatever reads it next. If you want to see the editor, the free web app opens any folder of .md files.