Canadian Technology Magazine is all about keeping up with useful technology, not merely collecting another pile of bookmarks that nobody ever opens again. A personal “second brain” is one of the most practical uses of modern AI: a local knowledge system that collects what matters, organizes it, connects it, and gives useful answers when you actually need them.
The core idea is simple. Your biological brain is excellent at thinking, making connections, and having ideas. It is not particularly reliable at remembering every deadline, article, meeting note, business insight, research paper, saved post, or random thought that lands in your life.
A second brain gives that information somewhere dependable to live. Large language models can now act like the librarian that earlier note-taking systems were missing. They can read source material, create summaries, link related topics, maintain living notes, and help turn a growing pile of information into decisions and action.
This is not about building a giant digital junk drawer. It is about creating a personal Wikipedia that is relevant to your work, your projects, your business, and your life.
LLM Wiki aka “Second Brain”
The second brain concept has existed for a long time. It has appeared under different names, including personal knowledge management, Getting Things Done, and the LLM Wiki. The underlying problem has always been the same: how do you collect everything coming at you, organize it without spending your entire life organizing it, and retrieve it at the moment it becomes useful?
In 1945, Vannevar Bush described the Memex, an imagined desk-like system that could retain what a person read and create trails between related ideas. That dream was about 80 years early. The missing piece was a tireless system capable of maintaining the library for you.
Now, models such as Claude, ChatGPT, Gemini, Codex, and Grok can do much of that librarian work. They can process raw information, identify entities and concepts, update notes, cross-link sources, and answer questions from the knowledge you have accumulated.
For Canadian Technology Magazine, this matters because AI becomes more useful when it has context. Instead of asking a model a generic question on the open web, you can ask questions against your own material:
- What subjects are gaining momentum that I have not covered yet?
- Which deadlines are coming up, and what needs to happen first?
- What changed in a platform or industry over the past few months?
- Which content formats or business processes are underperforming?
- What did I say about this topic years ago, and what happened afterward?
The system does not have to be limited to media, research, or technology. It can hold project notes, receipts, client information, study material, health notes, household tasks, meeting summaries, business processes, and ideas that would otherwise disappear into the void.
The important distinction is this: saving information was never the hard part. Most people already have unread bookmarks, saved posts, folders full of files, and notes scattered across apps. Maintaining, connecting, and using the information is the hard part.
tour of the “Second Brain”
A mature second brain starts to look less like a folder of notes and more like a knowledge graph. Every note can become a node. The connections between notes reveal relationships between people, projects, organizations, ideas, papers, analytics, content, and tasks.
Imagine one cluster for a business. It could include customer notes, contracts, key deadlines, services, recurring issues, and internal procedures. Another cluster might cover AI research, connecting frontier labs, papers, people, concepts, and summaries. A third could document a publishing workflow, linking articles, topics, performance data, and future ideas.
That is where a system becomes useful instead of merely impressive. If you can ask, “What popular AI subjects have I not covered?” the answer should come from your actual archive. The system can compare previous work, topic clusters, current trends, performance data, and recent research rather than guessing.
Canadian Technology Magazine can use this same structure to keep technology coverage organized across news, artificial intelligence, cybersecurity, cloud services, software, and business technology. The value is not in creating a pretty graph for its own sake. The value is in making the connections queryable.
From scattered records to useful context
A second brain can include both manual and automatic inputs. Manual inputs are the things you deliberately add: a URL, a voice note, an idea, a meeting summary, or a task. Automatic inputs come from routines that collect the information you care about on a schedule.
For example, a system can track social-post performance, archive relevant posts, calculate the velocity of a trending topic, and compare account performance against benchmarks. With enough data, it can identify when an algorithm changed, which content formats are no longer working, and what current best practices may be.
This is not supposed to be a collection of “growth hacks.” It is more about not accidentally shooting yourself in the foot because a platform changed while you were busy doing everything else.
A useful system can also maintain a map of the tools, applications, APIs, subscriptions, skills, routines, and automations connected to it. That provides a practical overview for security, costs, troubleshooting, and general sanity. Once a personal tech stack starts turning into a branching empire, one visual source of truth becomes very helpful.
Obsidian
The foundation of this approach is Obsidian, a free, local-first note application. Local-first means the notes are files on your own computer, not records locked inside somebody else’s cloud platform.
Obsidian uses Markdown files, usually ending in .md. Markdown is just plain text with a few simple symbols for formatting. A hashtag can create a heading. Asterisks can create bold or italic text. Links can connect notes.
This is a huge advantage. If Obsidian disappeared tomorrow, the underlying notes would still be yours. They remain readable text files. You are not locked into one company, one AI model, or one application.
For Canadian Technology Magazine and any organization trying to build durable knowledge assets, that ownership matters. Software changes. AI providers change. Pricing changes. Plain text remains plain text.
How links create the knowledge graph
In Obsidian, double brackets create internal links between notes. If a note about an AI researcher mentions an organization, it can link directly to that organization’s note. If a paper relates to a concept, the paper can link to the concept. Do this repeatedly and the system becomes a network instead of a stack of isolated pages.
Obsidian’s graph view makes those links visible. Each dot represents a file, while the lines represent relationships. Over time, dense clusters emerge around the things you work on most.
Keep the folder structure shallow. Do not build a maze of dozens of nested subfolders. Deep structures are annoying for people and can also become a nightmare for language models. Organize folders by the role a file plays, not by every possible subject.
A simple starting structure could look like this:
- Inbox: quick captures waiting to be processed
- Raw: original, unchanged source material
- Wiki: AI-maintained summaries, concepts, entities, and linked knowledge
- Index: a starting point for major areas of the system
- Log: a running record of activity and updates
The topics themselves should live in the links. “OpenAI,” “cybersecurity,” “customer retention,” or “cloud backups” do not necessarily need their own folders. They can be connected nodes that appear wherever they are relevant.
Ingesting New Info
Ingestion is simply the process of collecting information and bringing it into the vault. It can be as simple as giving an AI assistant a link and telling it to ingest the page.
A capable model can then read the source, store the raw material, produce a summary, update related concept pages, create links to entities, and ripple the new information through the rest of the knowledge base.
For example, a significant paper on AI interpretability can be added as one link. From there, the system can connect it to related ideas, labs, researchers, and earlier papers. The raw source stays intact, while the wiki layer becomes more useful over time.
Canadian Technology Magazine can use this process to keep a living research archive without manually rewriting every note. The model does the repetitive filing work, while the human remains responsible for deciding what deserves attention and what conclusions to act on.
Protect the raw layer
The raw folder should contain immutable source documents. In plain English, do not let the system rewrite the originals. The raw layer is the receipt. It is what lets you verify a claim later.
The wiki layer is where interpretation happens. This includes summaries, entities, concepts, cross-links, and useful context. Keeping those layers separate helps prevent a polished AI summary from being confused with the original evidence.
A practical rulebook, often stored in a file such as CLAUDE.md, tells the AI how to behave. Think of it as an employee handbook for the librarian. It can define:
- Which folders hold source documents and which hold summaries
- How notes should be named
- When a new entity or concept page should be created
- How links should be added
- Which sources and routines should be processed automatically
- What information requires human review
You do not need to hand-build every folder or rule. An AI coding assistant can create the initial vault structure and documentation from a clear request. The point is not to perform clerical work for fun. The point is to get the system running.
Kanban Board
A second brain becomes much more powerful when it knows about your real work, not only industry news. This is where a Kanban board comes in.
A Kanban board moves projects through visible stages. A simple content workflow might include:
- Idea
- Research
- Preparation
- Recording or production
- Editing
- Approval
- Published
For business work, the stages may be different: incoming request, planning, client review, implementation, testing, delivery, and complete. For personal admin, it could be to do, in progress, waiting, and done.
Obsidian supports this through a Kanban plugin, where cards can be dragged across the board. This is extremely useful for sponsor workflows, projects with deadlines, and anything that can quietly fall apart because a key step was forgotten.
Canadian Technology Magazine can apply the same model to editorial planning. Ideas can be scored for relevance and interest, moved into research, connected to source material, and promoted through production when ready.
AI can automate parts of the flow. When an idea moves into research, the system can gather related notes, find previous coverage, create suggested hooks, identify important dates and claims, and prepare a fact sheet. It does not have to write every word for you. It can simply make sure you do not walk into a complex topic with six tabs open and no idea where the important numbers came from.
For deadline-sensitive work, notifications can be connected through email, text, Slack, or another service. The board becomes a visual control panel rather than another forgotten productivity app.
the final “output”
The most important part of the system is the output layer. Raw data enters. The wiki organizes and remembers. The output layer answers the only question that really matters: what do we actually do?
One name for this layer is the “doctrine.” It may sound slightly dramatic, but it is a useful way to think about it. The doctrine contains actionable, evidence-backed playbooks built from your own data.
For example, a social analytics system might produce an updated guide showing which formats are working, what changed after an algorithm shift, which topics are accelerating, and where current performance is weak. Every recommendation should trace back to collected data rather than vague vibes.
The system can be thought of as four layers:
- War room: always-on routines and agents that gather intelligence
- Wiki: the organized, linked memory of the system
- Doctrine: decisions, playbooks, and actionable conclusions
- Armory: future tools, projects, and improvements worth building
This creates an OODA loop: observe, orient, decide, and act. Observe means collecting data. Orient means organizing it and understanding the context. Decide means forming a strategy. Act means executing the strategy and recording what happened.
Then the results go back into the system. Over time, the knowledge base compounds not only because it contains more data, but because it learns what happened after decisions were made.
That is the real opportunity for Canadian Technology Magazine: not just an archive of technology information, but a durable process for turning information into better editorial, technical, and business decisions.
How to Build This
You can build a useful first version in an afternoon. It will not be perfect, and it does not need to be. The goal is to create a foundation that gets more valuable the longer it runs.
1. Choose a local-first vault
Install Obsidian and create a new vault. Keep it on your computer. Start with a simple structure such as Inbox, Raw, Wiki, Index, and Log.
2. Choose an AI librarian
Use an AI assistant that can work with files and follow instructions. Claude Code, ChatGPT, Codex, or another capable model can work, depending on your setup. A subscription is worth considering if the system becomes part of regular work, because this is where the heavy lifting happens.
3. Write the rules once
Create a clear rulebook describing how files should be handled. Explain that raw documents are not to be modified, that wiki notes should be updated and linked, and that the folder structure should remain flat. Ask the AI to help create this file if needed.
4. Add useful plugins later
Obsidian plugins such as DataView and Kanban are optional. DataView can build automatic tables from your notes. Kanban provides the drag-and-drop board. Do not let plugin hunting delay the main system. The basic vault works without them.
5. Feed it consistently
Add one useful link, note, or document each day. Send interesting articles to the inbox. Capture voice notes. Store meeting summaries. Add tasks and deadlines. Over time, individual dots become a web.
Automation can come later. Start manually, learn what belongs in the system, then build routines around the information that matters most. It could be professional research, business operations, schoolwork, personal administration, or technology reporting for Canadian Technology Magazine.
6. Ask useful questions
Once the vault has material, use it. Ask for missing topics, overdue tasks, connected research, changes over time, recurring problems, and actionable next steps. A second brain only earns its name when it helps you think and act.
There will be a bit of resistance at first. Learning a new workflow can feel awkward, and setting it up takes more effort than saving another link for later. Push through that initial friction. A local library maintained by AI is one of those systems that gets better with every useful thing you add.
FAQ
What is an LLM Wiki?
An LLM Wiki is a connected personal knowledge base where a large language model helps organize notes, sources, summaries, concepts, and links. It is another name for an AI-powered second brain.
Do I need to know how to code to build a second brain?
No. A basic system only requires Obsidian, a folder structure, Markdown notes, and an AI assistant that can help create and maintain the setup. Coding can be useful for advanced automation, but it is not required for the core workflow.
Why use Obsidian instead of a cloud-only note-taking app?
Obsidian stores notes as local Markdown files. You own the files, they remain readable outside the app, and the system is not dependent on a single vendor continuing to offer the same service.
What should go into the Raw folder?
The Raw folder should contain original source documents, links, transcripts, records, and other materials that should remain unchanged. This layer acts as the evidence behind summaries and recommendations.
Can a second brain help a business?
Yes. A business can use it to connect project information, customer notes, recurring procedures, research, deadlines, analytics, and actionable playbooks. The system is especially useful when important knowledge is spread across too many applications and people.
How does Canadian Technology Magazine benefit from a second brain?
Canadian Technology Magazine can use a second brain to maintain connected technology research, identify developing subjects, preserve source context, organize editorial workflows, and turn accumulated information into more useful reporting and decisions.
