The Ingestion Phase
The Documents panel serves as a centralized, localized repository for user-generated assets, document attachments, and external data sources. Instead of managing disjointed conversational files, this module functions as an integrated split-pane layout that feeds content directly into your local vector space to anchor downstream agent reasoning.
The Documents interface featuring folder navigation and interactive asset cards.
📥 Ingesting New Assets
Clicking the+ Add button in the top-right corner exposes five distinct data integration workflows tailored to how you manage information:
The dynamic Add menu detailing local, web, and cloud ingestion workflows.
- Drop files: Drag and drop flat files directly into the UI (supports
TXT,MD,CSV,JSON,XML, and more). - Sync folder: Connect and monitor whole directories, importing all internal assets simultaneously.
- Add from cloud storage: Connect third-party cloud files directly to your knowledge base (displays connection health states).
- Add from URL: Provide a web link (
https://...) to scrape and clean online text content into local vector structures. - Import from path: Directly paste an absolute path string from your host machine file system to pull the target asset into the vault.
- Daily Ingestion Tracker: Monitor your remaining host computing resources via the rolling volume limit indicator (e.g.,
4999/5000 today) at the bottom of the menu.
📂 Interface Layout Architecture
The interface uses a two-column split structure to organize massive data sets without visual clutter:1. Left-Pane Folder Tree Navigation
- Monitored Only Toggle: Filter your entire list instantly to focus strictly on automated background directory watch folders.
- Directory Nesting: Navigate structural hierarchies (such as
Local > Medical) accompanied by numeric badges counting the exact number of files stored inside that sector.
2. Right-Pane Document Cards
Individual files are surfaced as robust interactive asset cards detailing automated extraction parsing summaries:- Checkbox Selection: Check individual card bubbles to trigger global batch tools like the Trash icon for bulk asset pruning.
- Metadata Footprint: View the ingestion date alongside the absolute source pathway (e.g.,
/Users/.../Desktop/Medical). - Text Extraction Snippet: Review an inline preview of the parsed data stream (e.g., OCR or text segments extracted by the core engine) to verify index legibility.
- Categorical Smart Badges: Color-coded keyword and collection indicators (e.g.,
Medical,mri-right-knee,acl-grade-iii) are automatically bound to the asset card for rapid visual querying.
⚙️ Card-Level Target Actions
Each document card exposes standard layout tools in its upper-right margin to modify configuration parameters or open source nodes:- Move to Folder / Space (Folder Icon): Remap the structural position of the file across your active storage paths.
- Line Graph Utility Icon: Inspect granular data generation points, query metrics, or specific vector chunk structures.
- Vertical-Dots Options Menu: Access system functions to copy file data, trigger manual re-indexing loops, or completely purge the document from your local machine.
Dynamic Artifact Tracking: Files generated by your chat sessions (PDFs, spreadsheets, images) support per-session custom output paths. If you move or rename these sessions, the underlying platform dynamically relocates the assets and re-indexes their paths automatically—ensuring your local document links never break.
Next Step: Memory Space — Chat History
Now that your core documents are accounted for, proceed to configure and audit your timeline logs inside the Chat History panel.