> ## Documentation Index
> Fetch the complete documentation index at: https://docs.selah.so/llms.txt
> Use this file to discover all available pages before exploring further.
# Knowledge Graph: Explore your theological concepts
> Selah auto-extracts theological concepts from your notes and captures into a browsable graph of themes, persons, places, doctrines, and more.
As you write notes and captures, Selah reads your content and automatically surfaces the theological concepts buried in it — themes like *grace* or *covenant*, biblical persons like *Paul* or *Mary*, places like *Jerusalem*, doctrines like *justification*, and more. These concepts are gathered into your Knowledge graph, a living index of what you've been studying and how ideas connect across your work.
You don't tag or label anything manually. The graph grows on its own as you write.
## Types of concepts
Every concept in your graph has a type. Selah recognizes six:
A recurring theological or spiritual idea — for example, *redemption*, *suffering*, *covenant faithfulness*, *the kingdom of God*. Themes tend to have the most connections because they show up across many passages and notes.
A biblical or historical figure mentioned in your study — *David*, *Peter*, *Augustine*. Person concepts let you trace how one individual's story or teaching threads through your notes.
A geographic location with theological significance — *Egypt*, *Sinai*, *Antioch*. Helpful for seeing where your study of a region or city has taken you.
A formal theological teaching or category — *atonement*, *election*, *the Trinity*, *sanctification*. Doctrine concepts tend to link to multiple notes and captures where you've wrestled with a belief.
A significant biblical or historical event — *the Exodus*, *Pentecost*, *the Fall*. Events often bridge multiple persons, places, and themes.
A book of the Bible your study returns to frequently. Book concepts connect to notes, captures, and passages tied to that book.
## How concepts are built
Selah extracts concepts automatically after you save a note, create a capture, or import a file. You don't need to tag anything or run a manual process — it happens in the background.
If new concepts don't appear right away, click **Sync** on the Knowledge page to trigger extraction manually.
Concepts appear a few seconds after content is saved. If you've just imported several files, give it a moment and then hit **Sync** on the Knowledge page if needed.
## Browsing and filtering
Open the **Knowledge** page from the sidebar to see all your concepts in a grid. Use the filter tabs — **All**, **Theme**, **Person**, **Place**, **Doctrine**, **Event**, **Book** — to narrow the view to one type.
Each concept card shows:
* The concept name and type
* A short summary (auto-generated or written by you)
* The number of connections to other content
## Viewing a concept's detail page
Click any concept card to open its detail page. There you'll find:
* **Summary** — a brief description of what this concept means in the context of your study
* **Last updated** — a relative timestamp ("Updated 3 hours ago") showing when the concept itself was last touched, regardless of whether it has a summary yet
* **Connections** — every note, capture, prayer, or teaching linked to this concept, with a preview of the content and a link to view it
* **Related concepts** — other concepts in your graph that this one connects to, and the relationship between them (for example, *illustrates*, *supports*, *theme of*)
## Concept summaries
Each concept can have a summary — a short description of what that concept means across your study.
If no summary exists yet, you have two options on the concept's detail page:
* **Write one** — click "Write one" to open an editor and type your own summary (up to 300 characters)
* **Generate Summary** — let Selah generate a summary based on your connected notes and captures
You can edit a generated summary at any time by clicking on it.
### Personal context in generated summaries
When Selah regenerates a concept summary, it now incorporates what the [Companion has remembered](/features/companion#what-selah-remembers) about you. If memory facts mention the concept by name — for example, a recurring conviction about *grace* — those facts are pulled into the summary so it reflects how you've come to think about the concept, not just its general theological definition.
This happens automatically each time a summary is regenerated. To refresh a summary with your latest context, click **Generate Summary** again on the concept's detail page.
## Merging duplicate concepts
Sometimes the same idea surfaces under slightly different names — *Grace* and *God's grace*, for example. You can merge two concepts into one.
Click **Merge** in the top-right of the Knowledge page. The button appears when you have at least two concepts.
Click the two concepts you want to combine. The page shows a counter — *0/2 selected* — as you make your choices.
Once two are selected, Selah asks which one to keep. Click the concept whose name and connections you want to preserve. The other concept's connections are transferred to the winner, and the duplicate is removed.
Merging cannot be undone. The losing concept is permanently deleted, though all its connections move to the surviving concept.
## How the graph connects to the rest of Selah
Your knowledge graph is not just a browsable list — it actively powers the rest of the app:
* **Search** draws on your concepts when finding results, so searching for a theme can surface captures that discuss it even without an exact word match
* **Companion** uses your graph as context, so when you ask a question it can draw on the concepts you've been building and connect them to your conversation
The more you write in Selah, the richer your graph becomes. Even quick captures contribute — every saved thought is scanned for concepts.
## Related
* [Search](/features/search) — find content across your graph
* [Companion](/features/companion) — have a conversation that draws on your knowledge
* [Import](/features/import) — bring in existing study material to grow your graph faster
* [Capture](/features/capture) — quick captures also feed concept extraction