App Settings

Customize the application display and tools from the App Settings panel, located at the bottom of the sidebar. All settings persist between sessions.

Theme

Setting	Type	Default	Description
Theme	Dropdown	Default Dark	Changes the color theme of the application. After selecting a new theme, refresh the page to apply it.

Note

Theme changes require a page refresh to take full effect. Your processing stack and document data are not affected by the refresh.

Performance

Setting	Type	Default	Description
Max workers	Number input (1 — 32)	8	Maximum number of parallel processes used during batch processing. Higher values process faster but use more CPU and memory.
OCR DPI	Number input (72 — 600)	300	Resolution used when rendering pages for OCR. Higher values improve text recognition accuracy but slow down processing.

Magnifier

Setting	Type	Default	Description
Enable magnifier	Checkbox	On	Toggles the magnifier lens on the PDF preview. When enabled, hovering over the PDF shows a zoomed-in overlay.
Zoom level	Slider (1.5x — 10.0x)	2.0x	Controls how much the magnifier lens zooms into the document. Higher values show more detail in a smaller area. Only visible when magnifier is enabled.
Magnifier size	Slider (120px — 800px)	400px	Diameter of the magnifier lens in pixels. Larger values give a wider viewing area. Only visible when magnifier is enabled.
Magnifier shape	Dropdown	Circle	Shape of the magnifier lens. Options are Circle, Rectangle, and Rectangle (Clamped). Clamped keeps the rectangle within the PDF preview bounds. Only visible when magnifier is enabled.

The magnifier renders a high-resolution version of the PDF page behind the scenes so the zoomed view stays sharp. Hover over any part of the PDF preview to inspect fine details like small text, dollar amounts, or signatures.

Tip

If you find the magnifier distracting while scrolling through pages, disable it from App Settings. You can re-enable it any time without losing your zoom and size preferences.

Splitter

Setting	Type	Default	Description
Default processor	Dropdown (Invoice, Document, Auto)	Invoice	Sets which processing pipeline the splitter uses by default when sending split documents for processing.
Auto-rotate pages	Checkbox	On	Detects and corrects rotated pages when loading PDFs in the splitter.
Auto-exclude blank pages	Checkbox	On	Automatically excludes blank or near-empty pages when loading PDFs in the splitter.
Blank page sensitivity	Slider (0.01% — 2.00%)	0.30%	Controls how aggressively pages are detected as blank. Higher values flag more pages as blank. Only visible when auto-exclude blank pages is enabled.

View Mode

Setting	Type	Default	Description
Scroll view	Checkbox	Off	Displays all documents in a scrollable list instead of one at a time. When disabled, the default 2-column layout is used.
2 columns layout	Checkbox	Off	Arranges document cards side by side in scroll view. Only visible when scroll view is enabled.
Enable OCR embed	Checkbox	Off	Embeds an invisible OCR text layer into scanned PDFs so they become searchable. Increases processing time.
Data Source	Dropdown	Current Server	Selects which vector database is used for search and chat. Options are Current Server, Old + Current, and Old Server Only. Selecting Old + Current or Old Server Only shows additional controls to download and build the old server index.

Note

When using the Old Server data source, a local copy of the old server index must be downloaded first. Use the Download DB button that appears below the data source dropdown, then Build BM25 to enable keyword search on the old server index.

LLM Integration

Setting	Type	Default	Description
Enable LLM for	Multi-select pills (Invoice, Document, Splitter)	All enabled	Toggles LLM-assisted processing per module. When enabled, the system uses Gemini to refine matching results. Disable individual modules to fall back to rule-based matching only.
Google API Key	Password input	Empty	Your Google AI Studio API key for Gemini access. Required for LLM features to work.
Invoice Custom Instructions	Text area	Empty	Custom instructions passed to the LLM when matching invoice fields. Use this to guide the model for specific edge cases, e.g. “Prefer matching by street number over entity name.”
Document Custom Instructions	Text area	Empty	Custom instructions passed to the LLM when classifying documents. Use this to guide the model for specific edge cases, e.g. “If document has both amendment and consent language, prefer Amendment.”
Splitter Custom Instructions	Text area	Empty	Custom instructions passed to the LLM when detecting document boundaries in the splitter. Use this to guide split behavior, e.g. “Exhibits should always be split as separate documents.”