# Atlas: your knowledge base builds itself.

Source: https://www.onsecondbrain.com/guide/atlas  
Language: en  
Spanish version: https://www.onsecondbrain.com/guide/atlas.es.md

---

The Atlas is the graph of the people, organizations and projects in your work. You do not fill it in: it is populated by what you write in your notes, what is said in your meetings and what you put in your documents. Every entity ends up with an entry, an article that writes itself and its links to everything else, each one showing where it came from.

## What goes in

### What an entity is

An entity is anything you talk about often enough to deserve its own entry. There are five types, each with its colour and its shape in the graph — shape as well as colour, so it reads the same with colour blindness.

- **People** — Who you work with. Their entry gathers their tasks, the meetings they appeared in, the documents that name them and their history across your notes.
- **Organizations** — Clients, vendors, teams. The full view of a relationship on one screen.
- **Projects** — What you are building. The natural unit for looking at progress without opening a project manager.
- **AI topics · Hobby** — The themes of your meetings. The AI spots them on its own in every recap and merges the repeated ones overnight; you do not create them by hand.
- **Tags** — Everything else you want to cross-reference: products, processes, places, concepts.

![The topics the AI recognized, in the Atlas](https://www.onsecondbrain.com/screens/guide/topics.png)

_Each topic opens with everything that mentioned it, in order._

### How it fills up, without you filling it

The Atlas has no registration screen to complete: entities come in while you work.

- You write a name in the daily note and detection proposes it as an entity.
- In a recap, the entity picker creates the missing one with + Create «…» in the Atlas.
- In a document, you select a name and add it with + Person, + Project, + Organization or + Tag.
- You create it by hand with the panel's +, when you want to prepare the ground before it shows up.

Every mention adds material to the entity. That accumulation is what later gives the article, the relations and the signals something to be built from.

### Unconfirmed: what the Atlas assumes

Some names show up again and again in your notes and recaps without anyone having added them. The Atlas does not ignore them, and it does not take them for granted either: it records them as unconfirmed.

- One is born when the same name appears on different days: a single mention is not enough. Generic words and countries do not count.
- It looks more transparent than the rest, and fades as days of activity go by without anyone naming it. Nothing is deleted by date: if it comes up again, it lights up again.
- They are shown and hidden with the Unconfirmed switch at the bottom of the panel. When you search, they all appear.
- Its entry says up front that it is not real yet — Not confirmed · assumed by the Atlas — and offers two ways out: Make it real, or It is one that already exists…, which merges it into a confirmed entity and keeps its name as an alias. The Atlas suggests which ones it could be.
- In the graph they have their own layer, Named alongside: who it gets named together with.

> **An assumption tags nothing**
>
> Until you confirm it, an unconfirmed entity never tags a task, enters a board or becomes anyone's assignee. It is there so you can see it coming, not to decide for you. For now they show on the computer; the phone shows confirmed entities only.

## What it builds

### The three layers of the graph

If every link is drawn the same way, you cannot tell what a person asserted from what a machine inferred. Here each layer is drawn differently.

1. **Declared relation** — A person stated it: client, team, partner, reports to. It is the strongest and it wins any conflict, because there is someone standing behind it.
2. **Wiki link** — One entity's article names another. It is a real, verifiable link — it is written down — but nobody declared it as a relation: it is inferred from the text.
3. **Semantic affinity** — Nobody wrote it: they keep showing up together in meetings, tasks and documents. This is the layer that discovers things, which is why it is drawn faintest: it is a lead, not a fact.

In the graph, hovering a node lights up its neighbourhood and dims the rest, so you can read what depends on whom without getting lost in the mesh.

![The workspace's knowledge graph](https://www.onsecondbrain.com/screens/guide/graph.png)

_The workspace graph. Color and shape carry the type — person, organization, project — and every line is a relationship that came out of your own material._

### What the Atlas infers by looking at content

Two signals, and both look at the content rather than the names. That is why they find things a name search does not.

- **Suggested relations** — Two entities that appear together more than chance would explain, or that the AI sees as linked, show up under Suggested, in the Relations tab of the entry. ✓ saves the relation; ✕ stops suggesting it.
- **Similar** — It compares what each entity's material is about, not what it is called, and shows the closest ones with their % match. That is how duplicates the name does not betray come up. For now, on the phone.

### The wiki that writes itself

Every entity has an article generated from all the material it appears in: your notes, your recaps, your documents, its tasks.

- Every claim carries the citation of its source: the note or meeting it comes from, one click away.
- The WhatsApp and Telegram assistant reuses it: when you ask it something about a client, the answer starts from their article. See WhatsApp and Telegram.
- It rewrites itself when new material comes in: every night the articles of the entities that changed are rebuilt. And if you do not want to wait, Regenerate rebuilds it on the spot.

### An entity's entry

It is the screen you open before a client meeting, and it replaces ten minutes of searching across three different apps. Everything about the entity sits in tabs:

- Wiki — its article. If the entity lives in more than one of your spaces, it carries the Also in… badge and opens with Across all your spaces: what the spaces share and what each one says.
- Relations — who it connects to, grouped by type, its Suggested ones and the Decisions that touch it. Beside them, the Map: hover a neighbour and the rest dims, and Open in the graph takes you to the whole graph.
- Tasks — its tasks and its milestones. You can mark them done right there.
- Folder — projects only: their documents, canvases, recaps, clips and files, with a search box and a filter by type. What merely names the project stays out: that goes in the History.
- History — everything that touched it, month by month and with its source: tasks, decisions from a recap, documents, daily notes.

Your own entry has one more tab, You: what is understood about how you work. See Secondbrain Me.

![A project's entry in the Atlas, tab by tab](https://www.onsecondbrain.com/screens/loops/ficha.jpg)

_A project in two spaces: Wiki, the relations map lighting up neighbour by neighbour, Tasks, Folder and History._

## Using it and keeping it clean

### Recaps and reports: what the Atlas gives back

The Atlas is not only somewhere to look: it is the structure that makes automatic summaries say something.

- Every recap links its entities when it finishes, so a client's entry fills up with every meeting you have.
- Period reports — from a day to a year — group what you did by project, client or person, not chronologically. That is why they are readable: they tell you what your week was about, not the order it happened in. See The digest.

### Merge, clean and maintain

A graph that grows on its own gets dirty on its own. These are the tools that keep it usable; almost all of them live on the computer, and on the phone you rename and delete:

- Optimize the Atlas — the ✨ button at the bottom of the panel reviews the unconfirmed entities and proposes, in one batch, what to merge, split, confirm, delete or restore (the ones you deleted that keep being named). You review every change before applying it.
- Merge — from the entry: "Acme", "Acme Inc" and "acme" become one, with all their material and relations, and the name that goes away stays as an alias.
- Rename — on a person it changes just the name, unless you pick Change the name and rewrite the past. On the other types the new name is written everywhere and the old one stays as an alias.
- Pin to top — up to five entities stay at the top of the list, outside the scroll, the filters and the search.
- Auto-merge duplicates and Purge unused tags — in the workspace settings, Advanced tab. The first joins the ones with the same name or matching another's alias; the second deletes in one go the ones that appear in no daily note, without going through the trash.
- Link a person to their account — if they use Secondbrain, you invite them from their entry and what concerns them reaches them. See Secondbrain Teams.
- Trash and history — at the bottom of the panel. Whatever you deleted is restored from the trash, links included. And the wiki history shows every version of the article: when it changed and who changed it, you, the automatic process or an agent over MCP.

### Keep reading

- [Memory, docs and canvas](https://www.onsecondbrain.com/guide/memory) — Documents, live collaboration, canvases, files and AI.
- [The digest and reports](https://www.onsecondbrain.com/guide/digest) — The report of the period, and the email that arrives on its own.
- [Meeting recaps](https://www.onsecondbrain.com/guide/recaps) — Record audio and screen, transcribe and pull the action items.