How to build a second brain you will still use next year

"Second brain" is Tiago Forte's name for a personal, digital store of notes, highlights and ideas that you can search and build from, so that your head is free for thinking rather than remembering ^[1]. The book is good at describing the system and silent on two questions that decide whether you will still be using it in a year: what memory research actually says about offloading your knowledge into software, and why so many second brains become write-only archives within a few months. This guide answers both, separates the parts of the method that rest on evidence from the parts that are folklore, and shows how AI search changes the design. For the tool decision itself, see our personal knowledge management tools comparison; for the older method that the second brain borrows from, see Zettelkasten.

The method: CODE and PARA

Forte's system has two parts ^[1].

CODE is the workflow: Capture what resonates (highlights, quotes, ideas, screenshots) with low friction; Organize by actionability rather than topic; Distill notes over time so their core stands out, through what he calls progressive summarization (bold the key lines, then highlight the best of the bold, then write a short summary on top); Express by turning the material into output, on the principle that the system exists to produce work, not to hold it.

PARA is the folder structure: Projects (things with a goal and a deadline), Areas (ongoing responsibilities with a standard to maintain), Resources (topics of interest), Archives (everything inactive). The rule is to file by where the note will be used next, not by what it is about, and to apply the same four folders in every app you use.

Two further ideas carry a lot of the method's weight: the slogan that "your mind is for having ideas, not holding them", and the claim that a well-kept second brain compounds, so that the value of what you captured rises over time.

What memory research supports

The method sits on top of a real and well-studied phenomenon: cognitive offloading, the use of external tools and actions to reduce the demands on memory and attention.

Claim in the method	What the research says	Grade
Offloading memory into notes is a sensible use of the mind	Offloading is adaptive and metacognitively driven: people offload more when the load is high and when they distrust their memory, and doing so frees capacity. It also carries a cost for unaided recall of what was offloaded, and people are not always well calibrated about when to do it [2]	A
Saving something frees you to take in what comes next	In three small lab experiments, saving one file before studying the next improved recall of the second, but only when saving was reliable [5]. Promising, small, not replicated at scale	C
"You remember where it is, not what it is" (the Google effect)	The original 2011 experiments found lower recall of trivia participants believed was saved [3]; the first experiment failed to replicate in a large preregistered project in 2018 [4]. Treat the effect as unproven in either direction	C
Resurfacing old highlights helps you retain them	Spaced review is one of the most robust findings in learning research, with the best gap growing as the retention interval grows [7]; practice testing and distributed practice are rated high-utility techniques [6]	A
Progressive summarization (layered highlighting) improves understanding	Highlighting and rereading are rated low-utility learning techniques [6]. The one layer of the method that does fit the learning research is the last: writing a summary in your own words is generative and resembles the elaboration and self-explanation techniques the review rates more highly	B
Information overload is the problem a second brain solves	Overload is conceptually rich and empirically thin: reviews of the literature find many descriptions and few measures [10]. Vendor surveys describe the cost of searching for information, about a quarter of the week in Atlassian's 2025 survey of 12,000 knowledge workers [11], but that is a retrieval failure inside organisations, not a limit of memory	B

So the evidence supports two things: offloading is a reasonable strategy with a price, and spaced retrieval of what you stored helps you keep it. It does not support the stronger slogans. "Your mind is for having ideas, not holding them" is a design preference; the offloading literature says you will remember less of what you hand over ^[2], which is fine if you will look it up and a loss if you will not. "The system compounds" has no test behind it, and the mechanism analysis below suggests the opposite is the default.

Why second brains die, and the fix

We do not yet have usage data of our own on template abandonment; what follows is a mechanism analysis built from the habit and memory research, and we will replace it with our own figures when the starter template has been in use long enough to count.

The cost comes first and the benefit comes later. Capture is cheap and immediately satisfying. Organising, distilling and reviewing are costs paid now for a return that arrives only when a future project needs the note. In habit-formation studies, the median time for a new daily behaviour to become automatic was around two months, with individual ranges from under three weeks to the better part of a year B^[8][9]. A second brain therefore has to survive roughly two months of effort with little visible payoff, on willpower alone, and the behaviours that need to become automatic (weekly review, filing by actionability) are the least rewarding ones.

Capture without retrieval is a write-only database. Every capture tool makes saving easier; almost none makes retrieval a habit. The result is a store that grows, is rarely opened, and is eventually abandoned because opening it has become a reminder of a backlog. The learning research is blunt about this: storing is not learning, retrieving is ^[6].

The structure answers the wrong question. PARA asks "where will I use this?", which is a good question on the day you file and an unanswerable one for most reading. Notes that do not fit a project end up in Resources, which is the folder nobody revisits.

Tool churn resets everything. Each migration between apps costs weeks and loses metadata. People who change tools yearly never reach the point where the store is large enough to be useful.

The fix follows from the diagnosis: design for retrieval first.

1. Start from outputs, not inputs. Name the two or three recurring things you produce (a weekly report, client proposals, a newsletter) and build the store around them. A note that cannot be connected to one of them is a candidate for not capturing.
2. Make retrieval the habit, not capture. Open the store at the start of each piece of work and search it before you search the web. Count uses, not notes.
3. Review for fifteen minutes a week, no more. Move or delete, do not reorganise. The review exists so the store stays trusted, which is the condition under which offloading pays ^[2][5].
4. Resurface deliberately. If you want to remember something rather than merely find it, schedule its reappearance at growing intervals ^[7]. Tools such as Readwise do this automatically for highlights.
5. Cap the capture. A daily limit on saved items forces the "will I use this?" question to the moment of capture, where it belongs.
6. Keep everything exportable. Plain text or Markdown files survive tool changes; proprietary databases do not.

Building it with AI retrieval

PARA exists because folder-based tools make you decide where a note lives. Semantic search over your notes, now built into Notion, available through plugins in Obsidian, and the whole premise of tools such as NotebookLM and Mem, weakens that constraint: a note can be found by meaning without being filed. Three design consequences follow.

• Under-organise. With reliable search, Projects and Archives are the two folders that still earn their keep, because they track what is active. Areas and Resources can collapse into tags or disappear.
• Mark your own words. AI summaries will enter the store, and in retrieval they look exactly like your notes. The automation literature is consistent that people over-trust automated output and under-check it, an effect that training and accountability reduce but do not remove ^[12][13]. Keep a visible distinction between what you wrote, what you quoted and what a model generated, and keep the source and date on every note.
• Keep a plain-text export even if the AI layer is excellent. The retrieval layer is now the most valuable and the most replaceable part of the system; the notes are the asset.

Interest in exactly this combination is where demand is moving. US searches for "second brain" run at about 3,600 a month, up 132 percent year on year although roughly flat over three years, while "second brain ai" (720 a month) grew 829 percent in a year (DataForSEO, retrieved 2026-11-09) ^[14]. Germany shows 1,000 monthly searches, up 307 percent, and the UK 590, up 126 percent; German readers get a localised version of this page.

Tool choices

The tool matters less than the retrieval habit, but it does set the ceiling on export, search and cost. Prices are from the vendors' pricing pages as recorded in our catalogue on 2026-11-09; where no affiliate programme exists, the tool is listed on equal terms. Our AI note-taking apps comparison covers the AI layer in detail.

Tool	Fits the method when	Entry price (checked 2026-11-09)	Export	Affiliate programme
Notion	You want databases, templates and AI search in one place and share with others	Free plan; Plus USD 10 per month billed annually, USD 12 monthly	Markdown, CSV, HTML, PDF	Yes
Obsidian	You want local Markdown files that outlive any app, with plugins for semantic search	Free for personal use; Sync USD 4 per month billed annually	Files are already plain Markdown	None
Readwise	Your inputs are books, articles and highlights and you want spaced resurfacing [7]	Full plan USD 9.99 per month billed annually; Highlights only USD 7.99	Export to Notion, Obsidian, Markdown	Yes
Evernote	You have a decade of notes there already	Free plan (50 notes); Starter USD 8.25 per month billed annually	ENEX, HTML	Yes
Apple Notes, Google Keep, a folder of text files	You want zero cost and zero setup and will rely on search	Included with the platform	Varies; text files are the safest	None

If you are starting from nothing, a folder of Markdown files (in Obsidian or any editor) plus one resurfacing tool is the cheapest system that satisfies every rule above.

Maintenance routine

Cadence	Action	Time
Each capture	Add source and date; ask which output it serves; skip it if none	10 seconds
Daily	Open the store before starting work; search before you search the web	2 minutes
Weekly	Review new notes: move to a project, tag, or delete. Count how many notes you used this week	15 minutes
Monthly	Archive finished projects; export a plain-text backup	20 minutes
Quarterly	Delete Resources you have not opened; check that the store still answers your three outputs	30 minutes

What to do next

Get the second brain starter template and run it for the two months the habit research says it needs ^[8][9]: a Markdown folder structure with Projects and Archive only, a capture note with source and date fields, a weekly-review checklist and a use counter. Then use the personal AI operating system builder to decide which tool holds it and which AI layer searches it, based on your inputs, outputs and privacy constraints rather than on what is popular this quarter.

Frequently asked questions

What is a second brain?

Tiago Forte's term for a personal digital store of notes, highlights and ideas, organised so you can find and build from them, with the aim of freeing your head for thinking rather than remembering ^[1]. The method has a workflow (CODE: capture, organise, distill, express) and a folder structure (PARA: projects, areas, resources, archives).

Is the second brain method backed by science?

Partly. Offloading memory into external tools is well studied: it is adaptive and frees capacity, at the cost of remembering less of what you offload ^[2]. Spaced resurfacing of saved material is strongly supported for retention ^[7]. The slogans are not: the claim that the system compounds has no test behind it, and progressive summarization relies on highlighting, which learning research rates as low-utility ^[6].

What is the PARA method?

A four-folder structure: Projects (goal and deadline), Areas (ongoing responsibilities), Resources (topics of interest) and Archives (inactive). Notes are filed by where they will be used next, not by topic, and the same four folders are used in every app. With semantic search over your notes, only Projects and Archives still do essential work.

Why do people abandon their second brain?

Capture is cheap and immediately rewarding; organising and reviewing are costs paid now for a benefit that arrives later. New routines take a median of about two months to become automatic, with a very wide range ^[8][9], and a store that is written to but never read becomes a backlog people avoid. The fix is to make retrieval, not capture, the habit.

Do I need Notion or Obsidian to build a second brain?

No. A folder of Markdown files satisfies every requirement, survives tool changes and costs nothing. Notion (free plan; Plus USD 10 per month billed annually) suits people who want databases and sharing; Obsidian (free) suits people who want local files with plugins; Readwise (from USD 7.99 per month billed annually) adds spaced resurfacing of highlights. Prices as recorded on 2026-11-09.

How does AI change the second brain?

Semantic search finds notes by meaning, so you can organise less and capture more. The risk is trust: people tend to over-rely on automated output and under-check it ^[12][13]. Keep what you wrote, what you quoted and what a model generated visibly distinct, keep sources and dates on every note, and keep a plain-text export.

Sources and evidence

1. Building a Second Brain: A Proven Method to Organize Your Digital Life and Unlock Your Creative Potential — Forte, T. (2022), book C (retrieved 2026-11-09)
2. Cognitive Offloading — Risko, E. F.; Gilbert, S. J. (2016), review A doi.org/10.1016/j.tics.2016.07.002 (retrieved 2026-11-09)
3. Google Effects on Memory: Cognitive Consequences of Having Information at Our Fingertips — Sparrow, B.; Liu, J.; Wegner, D. M. (2011), RCT C doi.org/10.1126/science.1207745 (retrieved 2026-11-09)
4. Evaluating the replicability of social science experiments in Nature and Science between 2010 and 2015 — Camerer, C. F.; Dreber, A.; Holzmeister, F.; et al. (2018), RCT A doi.org/10.1038/s41562-018-0399-z (retrieved 2026-11-09)
5. Saving-Enhanced Memory: The Benefits of Saving on the Learning and Remembering of New Information — Storm, B. C.; Stone, S. M. (2015), RCT C doi.org/10.1177/0956797614559285 (retrieved 2026-11-09)
6. Improving Students' Learning With Effective Learning Techniques: Promising Directions From Cognitive and Educational Psychology — Dunlosky, J.; Rawson, K. A.; Marsh, E. J.; Nathan, M. J.; Willingham, D. T. (2013), systematic review A doi.org/10.1177/1529100612453266 (retrieved 2026-11-09)
7. Distributed practice in verbal recall tasks: A review and quantitative synthesis — Cepeda, N. J.; Pashler, H.; Vul, E.; Wixted, J. T.; Rohrer, D. (2006), meta-analysis A doi.org/10.1037/0033-2909.132.3.354 (retrieved 2026-11-09)
8. How are habits formed: Modelling habit formation in the real world — Lally, P.; van Jaarsveld, C. H. M.; Potts, H. W. W.; Wardle, J. (2010), observational B doi.org/10.1002/ejsp.674 (retrieved 2026-11-09)
9. Time to Form a Habit: A Systematic Review and Meta-Analysis of Health Behaviour Habit Formation and Its Determinants — Singh, B.; Murphy, A.; Maher, C.; Smith, A. E. (2024), meta-analysis B doi.org/10.3390/healthcare12232488 (retrieved 2026-11-09)
10. The Concept of Information Overload: A Review of Literature from Organization Science, Accounting, Marketing, MIS, and Related Disciplines — Eppler, M. J.; Mengis, J. (2004), review B doi.org/10.1080/01972240490507974 (retrieved 2026-11-09)
11. State of Teams 2025 — Atlassian (2025), report B www.atlassian.com/blog/state-of-teams-2025 (retrieved 2026-11-09)
12. Complacency and Bias in Human Use of Automation: An Attentional Integration — Parasuraman, R.; Manzey, D. H. (2010), review A doi.org/10.1177/0018720810376055 (retrieved 2026-11-09)
13. Automation bias: a systematic review of frequency, effect mediators, and mitigators — Goddard, K.; Roudsari, A.; Wyatt, J. C. (2012), systematic review A doi.org/10.1136/amiajnl-2011-000089 (retrieved 2026-11-09)
14. Keyword database: DataForSEO Labs keyword overview, related keywords and historical search volume, US/UK/DE (data/research/keywords) — Productivity Signal research (2026), dataset A dataforseo.com/apis/dataforseo-labs-api (retrieved 2026-11-09)