How to Configure Super Brain Workspaces in Elephas

Workspace settings

Each Super Brain has configuration options that control how it behaves, what model it uses for chat, custom instructions, and which files to ignore from indexing. Open any Super Brain and click Edit (or the pencil icon) to see the Brain Details panel.

Brain Details panel

Brain name and description

Brain name: A short label for your workspace
Description: A short note about what you store in this Brain. A good description helps Elephas pull the right information when you ask a question

Auto-sync (Pro)

Auto-sync keeps your Brain fresh as your files and connected integrations change. It automatically re-indexes every 4 hours. Auto-sync is a Pro feature, so you will need to upgrade if you are on the free tier.

Smart Redaction

Choose how this Brain handles personal information when sending queries to cloud AI:

Inherit the global Smart Redaction setting
Always anonymize before cloud calls
Disable redaction for this Brain

Per-Brain Smart Redaction settings require Pro Plus. See the Sensitive Data Protection article for how Smart Redaction works.

Chat model

Pick which AI model this Brain uses to write its answers. The model controls how responses are written, not what is remembered. Cloud and offline models are both available depending on your provider configuration.

Indexing (embedding) model

Separately from the chat model, a Brain has an indexing model that turns your files into searchable embeddings. Options include a Small model (the default, text-embedding-3-small), a Large model for highest accuracy (requires your own OpenAI key), and fully offline models. You can set a default for all Brains in Model Settings, or override it per Brain here. Changing the indexing model requires re-indexing the Brain.

Custom instructions

Set custom instructions that apply to every chat in this Brain. Use this to set the tone, style, response format, or any rules the AI should follow when answering questions from this workspace.

You can even create an instruction-only Brain — one with custom instructions and no indexed documents at all. It's handy for content templates, decision frameworks, personas, structured-output helpers, or learning companions. Create one either during Create Brain > Advanced settings > Custom Instructions (skip adding documents), or later from the Knowledge Base panel.

Ignored Files

The Ignored Files section lives inside the Brain Details panel and lets you exclude specific files from the search index. The file remains in the workspace but is not used when answering questions.

Open the Brain Details panel
Scroll to the Ignored Files section
Click + Add and select the file(s) to ignore
Click Update Brain to save

Useful for excluding drafts, scratch files, or large media files you don't want surfaced in chat answers. You can also set ignored files during creation (Create Brain > Advanced settings > Ignored Files > Add).

Ignore rules support patterns, not just individual files:

*.jpg — all files with an extension
dir/* — everything directly inside a directory
**/name — recursive match anywhere in the tree
temp — a direct path/name string match

Open storage folder

The Brain Details panel also has an Open storage folder link at the bottom that opens the local folder where this Brain's indexed data is stored on your Mac.

Refreshing and Auto-sync

Connected sources (Notion, Apple Notes, Obsidian) don't auto-refresh unless Auto-sync is on. Your options:

Whole-Brain Auto-sync: the Auto-sync toggle in the bottom status bar of the Knowledge Base (Pro; re-indexes about every 4 hours)
Per-file/folder Auto-sync: right-click a file or folder > Enable Auto Sync
Manual: open a connected source and click Refresh / Sync Now

Note the difference: with Auto Sync off, Sync Now only re-checks files that are already indexed (new files in a folder are not picked up). With Auto Sync on, it also detects and indexes new files in indexed folders.

Checking or recovering a Brain's files

If a Brain stops showing up in the app, or you want to confirm its data is safely on disk, you can inspect the storage folder. When using iCloud storage (Preferences → General), Brain indexes live under:

~/Library/Mobile Documents/iCloud~Elephas/Documents/brains/

Each Brain is stored as a folder with a flattened name (for example aa1_faiss_index_elephas) containing two files: a .json file and a .faiss index. If both are present, the Brain's data is intact.

Backing up: zip the whole brains folder, for example in Terminal:

cd ~/Library/Mobile Documents/iCloud~Elephas/Documents/
zip -r brains.zip brains
mv brains.zip ~/Desktop/

To restore, unzip the folder back and create Brains with the same names so Elephas re-adopts them. Note a restored Brain may not list its files in the grid/list view, but search still works — this also works as a manual way to move a Brain between Macs.

Recovering a Brain that isn't showing in the app: create a new Brain with the exact same name as that folder. Elephas re-adopts the existing folder contents. Even if you re-add all the source files afterward, current versions only re-index the parts that actually changed, so it is quick.

Sharing a Brain

Use the Share button at the top-right of a Brain to generate a read-only link, shared securely via Apple Cloud. Recipients need Elephas installed. They can chat with the Brain but cannot see the underlying files or add their own. Updates you make auto-sync to recipients; if you stop sharing, they keep the last synced copy but receive no further updates.

Deleting a Brain

Click the red Delete button at the bottom-left of the Brain Details panel to delete a Brain. This removes all indexed data (the search index, text chunks, and metadata) from your local storage. Your original files on disk are not affected. This action cannot be undone.

Workspace Configuration