# The framework: write it once, let the rest follow.

Source: https://www.secondbrain.media/guide/framework  
Language: en  
Spanish version: https://www.secondbrain.media/guide/framework.es.md

---

secondbrain is not a note-taking app with AI bolted on: it is an organization method with one rule — what you already wrote is never written again. This page is laid out like the product flow: capture, plan and know. Each stage leans on the school that fits it — GTD, Eisenhower and PARA — and at the end come the research that holds it all up, when the method has to tell you something does not fit, and the order worth adopting it in.

## The flow, end to end

Everything that comes in lands in one place, and the rest is derived from there. You write, record, send a message or take a photo; out comes a planned week and knowledge that accumulates on its own.

_The four entry points are alternatives to each other and happen every day; what is on the right is a sequence. The three stages below are the three sections that follow._

- **Capture** — Recaps, Daily, Chat and Camera. Four doors into the same place, and none of them asks you to classify anything. It leans on GTD: capturing and clarifying are two different acts.
- **Plan** — Tasks and Calendar. What you committed to and when you will do it. It leans on Eisenhower, without asking you for the matrix: the quadrants come out of data you already entered.
- **Knowledge** — Memory, Atlas and Digest. What is written, who is who, and what you did coming back to you. The structure comes out of the text: here you do not do the filing.

> **The rule that orders everything**
>
> No data is entered twice. If a task came out of a meeting, the meeting is its origin and the link stays. If a person showed up in Tuesday's note, their Atlas card already recorded it. There is no screen for entering the same thing again: that function does not exist.

## Capture · what GTD got right

The first stage has one requirement: that writing costs nothing. That is why there are four doors and none of them asks you for a format.

- **Recaps** — The meeting gets recorded and leaves a transcript, a summary and action items without you taking a note. See Meeting recaps.
- **Daily** — One page per day, written in your own words. It is the main door and the one worth turning into a habit first. See The daily note.
- **Chat** — WhatsApp or Telegram: you send a message and it lands in the workspace. Good for whatever occurs to you away from the computer. See Messaging.
- **Camera** — A photo of a whiteboard or a handwritten page becomes text and comes in like everything else. See Mobile app.

What we owe GTD is this separation and nothing more: capturing and clarifying are two acts, and mixing them is what makes writing expensive. In secondbrain the cut is literal — while the note is open nothing exists; closing the day is what turns it into objects.

_To the left of the gate nothing exists in your workspace; to the right it does. That cut is what lets you write without watching yourself._

And there is a second, measured reason to write it down: getting information out of your head reduces the load you are carrying while you think — what Risko and Gilbert systematized in 2016 as cognitive offloading — and on top of that saving something improves what you learn next (Storm and Stone, 2015). The daily note is not an archive: it is what leaves your head free for what comes next.

> **What capture rests on**
>
> Two measured things, and each one decides a piece. Cognitive offloading — Risko and Gilbert, 2016; Storm and Stone, 2015 — is the reason the daily note is the center and the reason writing has no format.
>
> And the Zeigarnik effect, corrected: an open loop stops circling once you plan it, not once you finish it (Masicampo and Baumeister, 2011). Hence the day closing at a moment of its own, instead of an inbox reviewed whenever there is time.
>
> From GTD we keep the vocabulary and the one distinction that holds up — capturing and clarifying are two acts; the rest of this stage comes out of those two lines.

_What gets captured does not travel alone: the task that comes out of a line carries the whole branch it hung from, with its people and its project. That is what removes the need to enter the context again._

## Plan · Eisenhower, without the matrix

What you captured turns into two things: Tasks, which is what you committed to, and Calendar, which is when you will do it. The Eisenhower matrix names the four decisions in there well; what changes is that we ask you for none of the four by hand.

_The four quadrants and where each one lives. None of them asks you to classify: they come out of the date, of who executes, and of the quota you picked on Monday._

| Decision | The quadrant | Where it lives |
| --- | --- | --- |
| Do | Important and urgent. It goes now and pushes the rest aside. | In Tasks the task carries ❗ when it is due today, tomorrow or the day after, and the Planner places it inside your Peak. Anything already overdue carries a clock instead: it is not pulled forward, it is rescheduled. |
| Schedule | Important, not urgent. The highest-impact work, and what always loses against the urgent. | This is the quadrant the Planner was built for: Crítico and Impacto are only placed inside the Peak, never in whatever gaps are left. See Calendar and planning. |
| Delegate | Urgent, but not important that you do it. Someone else executes; you keep track. | You mark who executes: Yo reserves their work block, Espero reserves 15 or 30 minutes of follow-up. The Atlas person gets assigned, and if they also use secondbrain they see it on their side. |
| Drop | Neither important nor urgent. It is not going to happen, and it takes up room anyway. | Three exits depending on how final it is: the crossed-out eye discards the segment before it exists, Backlog takes it out of the week without deleting it, and postponing moves its planning. See Tasks and detection. |

The difference that matters is not about screens, it is about timing. The classic matrix asks you to resist urgency with willpower, which is exactly where you are measured to lose: people pick the urgent task over the important one even knowing the important one pays off more, only because it carries an urgency marker (Zhu, Yang and Hsee, 2018). That is why the Peak is reserved on Monday, before the week starts asking for things: when the urgent shows up on Wednesday, the important already has its slot. It is not discipline, it is an occupied slot.

> **What planning rests on**
>
> Three measured things, and each one decides a piece. The mere urgency effect (Zhu, Yang and Hsee, 2018) decides that the peak is reserved before the week starts. Little's Law decides that there is a quota at all. Circadian rhythms — Horne and Östberg, 1976; Roenneberg's MCTQ — decide that the day is split into bands instead of eight identical hours.
>
> From Eisenhower we keep the vocabulary: the names of the four quadrants are good and we use them as they are. What orders the decisions is the list above, and the last two are what give the week its two numbers.

Planning well then comes down to two numbers. The first is how much fits: one Crítico, three Impacto, five Operativas. That is not a preference, it is Little's Law — cycle time = work in progress ÷ throughput —: the more things you have started at once, the longer each one takes to finish. What does not fit is not prioritized: it is postponed.

_Nine slots in the week. It is a work-in-progress limit, and what it buys is not tidiness but speed: with fewer things started, each one finishes sooner._

The second is where each thing lands. You do not perform evenly all day: chronotype is a measurable, stable individual difference, and the Planner asks you When do you perform best? so it can spread work over that shape instead of over eight identical hours. Hard work goes in the peak; light work, in the slump.

_The curve of the day and the six bands. Reserving the peak is not a preference: it is the answer to urgency winning on its own._

> **Kairos and kronos: why the day is not a grid**
>
> The Greeks had two words for time. Kronos is time you count: clock hours, all identical, one after another. Kairos is the opportune moment: not how much time there is, but which moment is the right one for something.
>
> Almost every calendar is pure kronos. An even grid, every hour identical to the last, and a task goes wherever there is a gap. The curve above is the opposite: the six bands say that nine in the morning and four in the afternoon are not the same hour for you, and the Peak is precisely your day's kairos — the stretch where hard work actually comes out, and which is therefore reserved before anyone asks for it.
>
> That is where the rule that governs the split comes from: Crítico and Impacto work is only placed inside the Peak. Not because there are more hours there, but because not every hour is worth the same.

And once something has a time and a place it stops being a good intention: if it is Tuesday at nine, then I write the proposal is literally an implementation intention, which Gollwitzer formulated in 1999 and the meta-analysis with Sheeran (2006, 94 studies) measured at d ≈ 0.65 — a medium-to-large effect and one of the best replicated there is. That is why scheduling pays off more than prioritizing.

## Knowledge · what accumulates on its own

The third stage is the one that builds itself, and the one that takes time: a year in, the Atlas holds things about your clients and your projects that you no longer remember writing.

- **Memory** — Everything written and recorded: documents, canvases, clips and files, searchable together. See Memory, docs and canvas.
- **Atlas** — People, organizations and projects with their links, built from what you wrote without you entering them. See Atlas and knowledge graph.
- **Digest** — Every so often it hands you back what you did, sorted by topic, without you asking. See The digest.

> **The unit of work is the day, not the file**
>
> Everything that accumulates here comes out of the one thing you do write: the daily note. People, organizations, projects and links are derived from that text, and the effort of filing is never asked of you — not on saving, not afterwards.
>
> The name makes the comparison with PARA inevitable, so: Tiago Forte's school organizes what you save — project, area, resource, archive — and its unit of work is the well-filed document. If you enjoy organizing your own archive, PARA is better than this; if organizing it is exactly what you never get around to, this is the opposite bet.

What accumulates is not only yours. A team's memory is not the sum of its members' memories: it is knowing who knows what — what Wegner called transactive memory in 1985 — and it is exactly the part that never gets written down anywhere and is lost whole every time someone leaves a project. It is what the Atlas draws.

And the Digest is not there to look nice. Amabile and Kramer analyzed close to 12,000 work diary entries and found one factor above the rest: seeing progress in something you care about is what sustains motivation, more than rewards or recognition. The digest is that feedback, and it is the piece that keeps you writing in month four.

_The summary of the week you did not write. It comes out of the notes, meetings and tasks of the period, with people and projects already linked._

## Why what you already use is not enough

So far we have been talking about ideas. These are products, and the question is fair: if you already have somewhere to write, a task list and something that records your meetings, what is missing? What is missing is any one of the three taking charge of the whole flow. Each solves half.

_Two axes: how much of the work it covers, and how much it organizes itself. The ones that cover everything ask you to sort it out; the ones that understand without being told cover a slice. The top-right quadrant is nearly empty, and that is where the bet is placed._

- **Notes, wikis and task managers** — They cover almost all of your work and give you somewhere to put things. Entering, classifying and relating is still your job — which is exactly the work that gets abandoned the day you are swamped.
- **Recorders, assistants and memories** — They understand one surface very well: the meeting, the chat, the reminder. They solve that slice without you explaining anything, but what they produce does not accumulate into anything that grows on its own.
- **The closest ones** — Mem auto-organizes notes, Limitless records everything and gives it back, Motion schedules your tasks for you. Each solves a stretch of the flow; none walks it end to end.

We estimated each product's position ourselves, it is not measured: this is a positioning map, not a feature comparison. Several of these do their thing very well, and at the edges they overlap.

## What it is made of: the foundations, and how to check them

GTD, Eisenhower and PARA are the vocabulary the flow is explained in, which is why they are up there. None of the three is a foundation: two are practice methods and the third is a filing scheme. What holds each decision up is this other list — experiments, meta-analyses and mathematics — and each one predicts something that should show up in your own workspace within a month.

| The foundation | What it holds up | How you check whether it works for you |
| --- | --- | --- |
| Cognitive offloading | That the daily note is the center, and that writing costs nothing. Risko and Gilbert, 2016 · Storm and Stone, 2015 | Writing stopped costing you and you write more than in week one. It is the one non-negotiable requirement: everything else feeds off it. |
| Zeigarnik, corrected | That closing the day exists: what is pending does not stop nagging when you finish it, it stops when you plan it. Masicampo and Baumeister, 2011 | You end the day with fewer things rattling around, even if you did not do more. If you do not notice it, you are probably closing the day without looking at what got marked. |
| Attention residue | Entering things once, and few long blocks instead of many short ones: the cost of switching is at the edge, not inside. Leroy, 2009 | Two one-hour blocks pay off less than one two-hour block. That is why Espero follow-ups are grouped. |
| Implementation intentions | That planning means putting a time and a place on things, not tidying a list. Gollwitzer, 1999 · Sheeran, 2006: d ≈ 0.65 across 94 studies | Compare it yourself: what proportion got completed of what went through the Planner versus what stayed loose in the list. |
| Little's Law | The nine-slot quota, which is a work-in-progress limit. Queueing theory | The time between a task appearing and closing goes down. If what went up is the volume, you raised the quota, not the speed. |
| Circadian rhythms | That the day is split into bands rather than eight identical hours. Horne and Östberg, 1976 · Roenneberg's MCTQ | Crítico work gets completed more inside the peak than outside it. If it completes the same in any band, your peak is set wrong — three clicks to change it. |
| The mere urgency effect | Reserving the peak before the week starts, instead of asking you to resist. Zhu, Yang and Hsee, 2018 | On Wednesday, when the urgent shows up, the important already has its slot. If it gets pushed aside anyway, you reserved the peak too late. |
| The progress principle | That the Digest exists, and that it is not optional in the method. Amabile and Kramer, across some 12,000 diary entries | Reading the digest gives you back the sense of having moved forward. If four in a row leave you indifferent, it has not become a habit yet. |
| Transactive memory | That the Atlas records who knows what, and where each link came from. Wegner, 1985 | On a team: when someone leaves a project, what they knew is still written down somewhere. |
| Automation and human factors | That the AI proposes and you confirm: an AI that creates on its own produces two failures at once, first you over-trust and then you switch it off. Parasuraman and Riley, 1997 | You are still reviewing what it marked. The day you stop looking, or the day you switch it off, the automation went too far. |

None of these checks are run for you by the tool: they are the questions worth asking, with your own data, after the first month. We write them down anyway because a method that cannot be proven wrong is not a method, it is a promise. And they hold for other people under other conditions: they are good reasons to build the method this way, not proof about you.

_The three mechanisms by which work comes apart, and the method's answer to each. The rows are not steps: they are three fronts happening at once._

## And a theory of stopping

Almost every productivity method is a theory of starting, and without a theory of stopping what is left is a machine for cramming more into the same week.

The effort-recovery model (Meijman and Mulder) says something simple and well measured: the cost of an effort is paid off with recovery, and if the recovery does not happen the cost does not disappear, it accumulates. Sabine Sonnentag's line of work on psychological detachment adds the uncomfortable detail: what predicts how you start on Monday is not how much you worked on Friday, but how much you managed not to think about work in between.

That is why the method has five ceilings, and none of them is a wellbeing feature:

- The peak is finite. Crítico and Impacto work is only placed inside the Peak, so the size of your peak — not the number of hours in the day — is the real ceiling of your week.
- The quota is a ceiling, not a target. Nine slots. Nobody wins by filling them, and filling them every week is the sign that the quota is set wrong.
- The days you do not work are data. Which days do you work? is not a courtesy question: outside those days the Planner places nothing.
- Closing the day closes the day. The note locks, and while a note is open the day stays open in your head. The gesture is right there for you to press —pressing it is what marks the end— and automatic closing is on by default, because a day that closed itself can be reopened and a week of open days cannot be recovered. It is switched off per workspace.
- The Pomodoro puts the ceiling inside the day. The other four work by week or by workday; this is the only one that cuts in while you are working. One focus block and one break, on a clock that keeps running with the app closed — in plain sight in the Mac menu bar or on your phone's lock screen — so the pause does not depend on you remembering.

> **And why the Pomodoro does not contradict "few long blocks"**
>
> The attention residue foundation says it is better to load context once and work in a few long blocks, because the cost of switching is at the edge. A clock that cuts in every 25 minutes looks like the opposite, and it is not: what it cuts is the rest, not the work. Inside a focus block you do not switch tasks — that is the point — and after the break you go back to the same thing. Residue is the price of changing subject, not of stopping.
>
> And if your unit of work is longer, the clock stretches: focus is configurable from 1 to 180 minutes and the break from 1 to 60. The 25 + 5 is a default, not a rule of the method. Someone who works in ninety-minute blocks sets ninety.
>
> What does belong to the method is the asymmetry: the break exists so that recovery happens inside the day instead of being saved up for the weekend — which is exactly what the effort-recovery model says does not work.

> **Why this is part of the method and not an extra**
>
> A system that only knows how to add work measures badly by construction: the benefit shows up this week and the cost appears two weeks later, somewhere else and without a label. By the time it shows, nobody attributes it to the method any more.
>
> It is not a promise of balance or a break counter. It is a constraint: if the framework cannot tell you that something does not fit, it is not prioritizing anything.

## What order to adopt it in

You do not need to switch everything on the first day, and the order is the order of the flow: capture first, plan next, and the knowledge builds itself in the meantime.

One warning before the numbers: in Lally et al. (2010) a new habit took a median of 66 days to become automatic, with a spread from 18 to over 250. "Week 1" is a label for order, not a deadline: each layer opens when the previous one stopped costing you.

1. **Week 1 — just the daily note** — One page per day and nothing else. No tasks, no projects, no planning. The only goal is for writing there to become automatic. If that does not happen, none of the rest will either.
2. **Week 2 — close the day** — Start pressing Close day when you finish work. You will see tasks appear in Tasks that you did not create. Discard the wrong ones: every discard teaches, and within days detection tunes to how you write.
3. **Week 3 — plan the week** — Answer When do you perform best? and Which days do you work? once, and press Planner on Monday morning. This is where the method starts giving time back instead of asking for it.
4. **Week 4 — record your meetings** — Switch on Recaps. It is the layer that adds the most volume: every meeting leaves a transcript, a summary and action items without you taking a note.
5. **After that — the Atlas and documents** — By then the Atlas has built itself from what you wrote. Only then is it worth going in to look at it, completing cards and writing in Docs what is worth keeping.
6. **Once it is a habit — open the channels** — Mobile, WhatsApp or Telegram, the Web Clipper and MCP. Opening them before you have the habit does not help: there is nothing inside yet.

_Week 1 is only this: one page per day, written the way you would have written it anyway. On the right, what the app already inferred without you entering anything separately._

> **What the framework will not give you**
>
> It will not tell you what to work on: you choose that, and the planner is designed to force you to choose rather than deciding for you. It does not replace thinking. And it does not work if you do not write: everything it returns comes out of what you gave it, so an empty note returns nothing.

## What needs a plan, and which one

The full method does not fit in the free plan, and it is better to know that before starting than halfway through week 3. Everywhere else in the guide, each feature that needs one carries its mark — · Hobby or · Pro — next to the title. This is the whole list, in one place.

| Feature | Minimum plan |
| --- | --- |
| Writing the daily note, documents, canvases and tasks by hand | Free |
| Recording and transcribing meetings | Free |
| Calendar, drag and drop and connected calendars | Free |
| The digest | Free — one per week |
| MCP and the Web Clipper | Free |
| Capturing through WhatsApp and Telegram (text) | Free |
| Transcription: 30 min per month · one workspace · 50 MB | Free |
| AI detection in the daily note — tasks, entities and closing the day | Hobby |
| Weekly planner — the quota and the energy-based layout | Hobby |
| AI deliverables | Hobby |
| AI summaries of recaps | Hobby |
| Generating and rewriting documents with AI | Hobby |
| Read aloud | Hobby |
| Atlas topics — the theme grouper | Hobby |
| Transcription: 20 h per month · a second workspace · 1 GB | Hobby |
| Screen understanding | Pro |
| Create from an image — a whiteboard photo or handwriting | Pro |
| Audio and images through WhatsApp and Telegram | Pro |
| Scheduled digest by email | Pro |
| Assistant through WhatsApp and Telegram — asking about your own material | Pro |
| Transcription: 100 h per month · up to ten workspaces · 10 GB | Pro |

Recording hours, storage and the number of workspaces are ceilings, not switches: you are warned as you approach them and what you already uploaded is never touched. Prices are on the plans page.

## The map: where everything lives

The pages that follow document each piece in detail, in the order of the flow: first what gets captured, then what gets planned, then what accumulates. At the end, the ways in and the settings reference.

- [The daily note](https://www.secondbrain.media/guide/dailynotes) — Bullets, inheritance, predictive typing and dates.
- [Tasks and detection](https://www.secondbrain.media/guide/tasks) — Where your tasks come from and how they group into deliverables.
- [Calendar and planning](https://www.secondbrain.media/guide/calendar) — The curve of the day, your peak window and the weekly quota.
- [Meeting recaps](https://www.secondbrain.media/guide/recaps) — Record audio and screen, transcribe and pull the action items.
- [Memory, docs and canvas](https://www.secondbrain.media/guide/memory) — Documents, live collaboration, canvases, files and AI.
- [Atlas and knowledge graph](https://www.secondbrain.media/guide/atlas) — People, projects and organizations, linked on their own.
- [The digest and reports](https://www.secondbrain.media/guide/digest) — The report of the period, and the email that arrives on its own.
- [Web and desktop app](https://www.secondbrain.media/guide/web) — The full workspace, in your browser or on your desktop.
- [Mobile app](https://www.secondbrain.media/guide/mobile) — Capture, record and listen from your phone.
- [WhatsApp and Telegram](https://www.secondbrain.media/guide/messaging) — Capture over chat, with text, audio or images.
- [Web Clipper](https://www.secondbrain.media/guide/clipper) — Save any page from your browser, already read and classified.
- [MCP and AI clients](https://www.secondbrain.media/guide/mcp) — Connect Claude or ChatGPT to your workspace.
- [Settings](https://www.secondbrain.media/guide/settings) — Both settings screens, tab by tab.