Explore Synapses
Browse the map of pages, facts and entities Brain built from your sources, and see how they connect.
Synapses is a map of what GuruSup Brain learned from your sources. Each dot is a node, such as a page Brain wrote, a fact or an entity. Lines between them are relations. Use it to see what Brain holds on a topic, what it links to, and where a fact came from.
Synapses is read-only. You can look, search, filter and follow links, but you can't change or delete anything from here.
Only admins can open Synapses. It sits under Knowledge in the sidebar. Members who try the address are sent back to their home page. Before the memory build starts, the page shows a preview with "Available once your memory is built", and admins see a Get started button.
What you can do with it
| You want to | Do this |
|---|---|
| Check what Brain knows about a customer or project | Search for the name, then select the node. |
| See where a fact came from | Select the fact. The page that asserts it is listed under Referenced by. |
| Find what a page mentions | Select the page and read References. |
| Spot topics that are thinly covered | Look for pages with few relations, or filter by a category. |
Read the map
The top left shows the count of what is on screen, for example 120 nodes / 340 relations. The map has two views, which you switch with the buttons at the top right:
| View | What you get |
|---|---|
| Canvas view | A 3D map. Drag to orbit, scroll to zoom. Double click a node to open its page. |
| List view | A list of nodes with their relations. It is the default on narrow screens. |
On a narrow screen the explorer panel with search and filters is hidden. Use a wider window to get it.
What the dots and lines are
Nodes
| Node | What it is |
|---|---|
| Page | A page of Brain's memory, written from your sources. It carries a link to where it came from when there is one. |
| Section | A heading inside a page. Sections start hidden. |
| Fact | A statement Brain took from a page. Its name is the first part of the statement. |
| Entity | A named thing Brain recognized in your content and can link across pages. |
Relations
| Relation | What it means |
|---|---|
| has section | A page contains this section. |
| mentions | A page mentions this entity. |
| links to | A page links to another page. |
| asserts | A page states this fact. |
| about | A fact is about this entity. |
Relations point from one node to another. In the panel, References are the ones leaving the node and Referenced by are the ones arriving.
Colors and size
Size follows how many relations a node has, so the most connected nodes stand out. Color follows the node's category when it has one. A node without a category takes a color from its kind.
| Category | Color |
|---|---|
| general | Gray |
| marketing | Pink |
| sales | Green |
| finance | Amber |
| product | Blue |
| engineering | Violet |
| customers | Cyan |
| hr | Lime |
| leadership | Red |
Any other category gets a color from a fixed set. Each category chip carries its color, and the same color appears on the node's row in the lists. In the canvas, names show on the 60 most connected nodes, and on the selected node and its neighbors. Turn all names off with Labels.
Find a node
Search
Type in Search node… in the panel on the left. Brain lists up to 50 matching nodes among those already loaded. It matches the node's name and kind, and puts the most connected first. Pick one to select it and fly to it on the map. If its layer was hidden, Brain turns it on.
Search the whole graph
If the node isn't loaded, press Enter or select Search «your text» across the whole graph. The map then shows results for that search, with Results for «your text» at the top. This search ignores the node limit and returns up to 400 nodes. Select Whole graph to go back.
If nothing matches, you'll see No loaded node matches.
Filter what you see
The panel on the left has these controls:
| Control | What it does |
|---|---|
| Layers | Turn on or off Pages, Entities, Facts and Sections. A layer only appears if the map has nodes of that kind, and shows how many. Sections starts off. |
| Simplify | Hides nodes with no relations and entities that only one other node points to. It starts on. |
| Labels | Show or hide node names on the map. |
| Categories | Show only nodes in the categories you pick. |
How they work together:
- Categories are multiple choice. A node stays if it matches any category you picked. Brain also keeps neighbors that have no category, such as entities, so the picture stays connected. A category chip disappears when its nodes are all on hidden layers.
- The category counts follow the layers you have on.
- The node you selected stays visible even when a filter would hide it.
- Reset filters turns every layer back to its default, clears the categories and turns Simplify off. It appears when your filters leave no node visible.
Below the filters, the panel lists the nodes on screen, most connected first, up to 250. It says "+N more…" for the rest. Select one to reveal it on the map.
At the top right you also have:
- a Node limit menu, which sets the maximum Brain loads.
- a Refresh button to reload the map.
Node limit
The options are 100, 200, 400, 800, 1500, 2000 and 10000 nodes. It starts at 400. The menu is off while the map is loading.
If you can't find what you need:
- Search the whole graph first. It doesn't depend on the node limit.
- Raise the node limit one step at a time. Larger limits load more and take longer to load and to draw.
- Narrow with Categories and Layers instead of showing everything.
- On a slow computer, switch to List view. It shows at most 500 nodes, and says how many are hidden. Narrow with filters or search to see the rest.
Open a node
Select a node on the map or in the list. The map narrows to that node and its connections, and the top left adds a line such as 12 in selection. A panel opens on the right showing:
- the node's name, its kind and its categories,
- Out, In and Total: how many relations leave the node, arrive at it, and both together,
- References: the nodes this one points to,
- Referenced by: the nodes that point to it.
Both lists group the connections by relation, up to 25 each, and say how many more there are. Select a node in either list to jump to it. If a node has none, the panel says No relations.
Use Clear selection at the top, or Close details on the panel, to go back to the full map.
Expand neighbors
Select Expand neighbors in the panel to load the nodes around this one into the map. The button shows Expanding… while it loads, then Neighbors loaded. Expanded nodes stay until the map reloads.
Open page
When a node has a link attached, the panel shows Open page. It opens that link in a new tab. Double clicking the node on the map does the same. It opens the link stored on that node, which is not always your original document. What Brain stores is its own memory of your sources. See Sources and citations.
List view
Each row shows a node's name, its kind and how many relations it has, for example 4 rel.. Open a row to see its relations, written as the relation name and the node it leads to. Select a relation to jump to that node. A node with none says No relations.
When something goes wrong
| You see | What to do |
|---|---|
| A spinner and "Computing layout…" | The map is loading. Larger limits take longer. |
| No graph data, with "The graph is built from your pages, memories and sources." | Connect a source and let it finish learning. See How Brain works. |
| "The current filters leave no node visible." | Select Reset filters. |
| "Could not load the graph." or "Could not search the graph." followed by a reason | Select Refresh, or try again in a minute. If it keeps failing, contact support. |
| "Could not expand the node." followed by a reason | Select the node and try Expand neighbors again. |
How fresh the map is
Brain keeps the map it loaded for 30 seconds, so switching pages and coming back within that time doesn't reload it. To see the latest, select Refresh. Any reload clears your selection and any neighbors you expanded, and so does changing the node limit.
The map lags behind your sources by however long Brain takes to learn new content. See Follow ingestion.
Related
- Sources and citations: what a page in Brain's memory is.
- How Brain works: how content becomes pages, facts and entities.
- Follow ingestion: check that new content has been learned.