Buying Guide
Best Second Brain Apps in 2026
The best second brain apps in 2026, from Obsidian and Notion to Tana and Capacities, and how to pick by what you need to recall.
The best second brain app for most people is Obsidian if you want to own your files, or Notion if you want structure without maintaining it yourself. Both are excellent, both have been for years, and the difference between them matters far less than the thing nobody asks before installing either: what are you actually trying to recall?
That question sorts this category better than any feature comparison, because every app below is built around ideas you will reread, and almost none of them are built around people you will meet again.
Sort by what you ask it, not what it stores
Two questions get asked of a second brain, and they need different shapes.
“What do I know about supply-chain risk” is a topic question. Every app here answers it well, because topics are what the whole category was designed around.
“What do I know about Marcus” is a person question, and it behaves differently. Its organizing key is a name, its useful moment is the five minutes before you see him, and the material is fragments rather than documents. Filed by topic it is unfindable. Filed by date it becomes a diary you never reopen.
Most people need both. Almost nobody has both, and the second one is usually the gap.
The landscape
The vertical axis is the honest predictor of whether a system survives. Every hour of schema you design up front is an hour you enjoyed, and a tax you pay on every note afterwards.
The contenders
| App | Model | Best for | Where it stops |
|---|---|---|---|
 Obsidian | Local markdown files with links | Owning your notes as plain files, forever | Setup and plugins are yours to maintain |
 Notion | Databases, pages and templates | Structure and sharing without building it | Databases drift toward upkeep work |
 Apple Notes | Plain notes with folders | Fast capture that is already on the phone | Search is literal, no linking of substance |
 Evernote | Notebooks and web clipping | Clipping and archiving reference material | Long-standing pricing and platform churn |
 Logseq | Local outliner with daily notes | Outline thinkers who want local files | Block-first structure suits some brains only |
 Reflect | Networked daily notes | Daily journaling with backlinks and AI | Opinionated shape, subscription only |
 Capacities | Typed objects rather than pages | Treating people and places as real objects | Object types need designing before they help |
 Tana | Outliner with supertags and fields | Power users who genuinely want a schema | The steepest learning curve here |
 Intriq | Notes attached to people | Recalling what someone told you, before you meet | Not a general notes app: people only |
The setup tax
The pattern that ends most second brain projects is not choosing wrong. It is choosing well and then over-building.
A schema you design in January describes the work you imagined in January. By March the work has changed, the schema has not, and every note now requires a small decision about where it does not quite fit. That decision is where capture dies.
The practical rule: start with the least structure you can tolerate, and add a field only after you have missed it three times. Tana and Capacities are genuinely powerful, and they reward people who already know the shape of their own thinking. They punish people using setup as a substitute for starting.
How to choose
- Do you want to own plain files? Obsidian, or Logseq if you think in outlines. Local markdown, yours in twenty years.
- Do you want structure you did not have to build? Notion. Templates cover most of what people build by hand.
- Do you mostly need fast capture on a phone? Apple Notes is genuinely enough, and it is already installed.
- Do you want typed objects, including people? Capacities, or Tana if you want full control of the schema.
- Is the thing you keep forgetting about people rather than topics? That is a different shape, and it is the next section.
The people-shaped gap
Every general system above can technically hold people. You make a page per person and link to it. This works for about six weeks.
It degrades for a reason worth naming: general systems put the burden of retrieval on you at exactly the wrong moment. You have to remember that the person page exists, remember to open it before a meeting, and remember to have written it after the last one. Three acts of memory to support a system whose job was to replace memory.
The fragments also do not want to be documents. What someone told you is four half-sentences, not a page: a job change, a worry, a promise you made, a name they mentioned. Writing that into a structured page feels like overkill, so it does not get written, so the person page stays empty and the system quietly proves it was never the right container.
Where Intriq fits
Intriq is a second brain for people only, and the narrowness is what makes the capture cheap enough to keep doing.
There is no notebook to choose, no tag to pick and no page to create. You write or speak a few fragments after seeing someone, and Intriq attaches them to that person and dates them. Filing happens after the writing rather than before it, which is the order that keeps notes getting written.
Recall runs the other way too. Before a meeting you get what that person told you last time, surfaced because you are about to see them rather than because you remembered to look. Its AI answers only from what you have saved and shows you the note it used, so asking what someone needed returns your own words rather than a plausible guess.
Notes are local-first with encrypted on-device snapshots, and Intriq deliberately does not enrich contacts or scrape public profiles. It runs on iPhone and in any browser at app.intriq.app.
It pairs with a general second brain rather than replacing one. Keep ideas, projects and sources in Obsidian or Notion, where they belong. Let the people layer live somewhere built for fragments and names.
Key takeaway: Pick a general second brain on the setup axis, since the least structure you can tolerate is the one still running in April, and Obsidian or Notion covers most people. Then check what happens to people in your system, because that is the material topic-shaped tools drop.
FAQ
What is the best second brain app?
Obsidian if you want to own plain local files, Notion if you want structure without building it. Both are strong general choices. The better question is what you need to recall, since both are organized around topics and neither handles people well.
Is Obsidian or Notion better for a second brain?
They optimize for different things. Obsidian gives you local markdown you control completely and expects you to assemble your own setup. Notion gives you databases and templates that work immediately and live in Notion’s cloud. Pick by whether file ownership or zero setup matters more.
Are second brain apps worth it?
Yes, when kept boring. The systems that pay off are the ones with almost no setup, because capture friction is what kills them. Elaborate schemas built before any real use are the most common way people spend a weekend and abandon the result.
What is the best free second brain app?
Obsidian is free for personal use, Logseq is free and open source, and Apple Notes is already on your phone. Notion has a free personal tier that covers most individual use.
Can a second brain app remember people?
Only with work. You can make a page per person in any of these, but general systems are topic-shaped, so people fragments have nowhere natural to land and retrieval depends on you remembering to look. Capacities and Tana model people as objects with setup, and a dedicated people layer does it without any.
How do I stop abandoning my second brain?
Reduce the number of decisions between having a thought and writing it. Start with one flat capture surface, add structure only after missing it repeatedly, and be suspicious of any afternoon spent configuring rather than capturing.
For the adjacent thinking, read AI second brain, second brain for relationships and Obsidian vs relationship memory.