Document Processing

Document processing converts uploaded (or imported) files into content the chat agent can retrieve and cite.

Pipeline (what you see)

1. Upload / import — File enters the workspace Library.
2. Extract — Text is pulled from the file; images/scans may use OCR.
3. Chunk — Long content is split into pieces sized for retrieval.
4. Index — Chunks become searchable for semantic retrieval in chat.
5. Ready — The document appears available in Library and via @ mentions.

Status is shown in the Documents experience while processing runs. Large or complex files take longer.

Supported file types

These are the main types AINexLayer accepts on upload (from the app’s supported-file list):

If a file is rejected, convert it to PDF or DOCX and try again.

Basic vs Premium processing

When you upload files, pick a Processing Mode in the upload dialog:

How pages are counted:

• PDF — one real PDF page = one billable page
• Word, PowerPoint, Excel, and similar — pages are estimated automatically
• Every processed file uses at least 1 page

Credit is checked before processing. If the balance is too low, the upload is paused until you add credit.

Ways content enters the Library

• Upload Files from Library or the chat + menu
• Cloud file imports via supported drive connectors (see External connectors)
• Built-in memory files such as workspace memory notes (when present in your Library)

Live connectors (for example Slack search) are different: they answer in real time and do not bulk-index your entire account into Documents.

Quality tips

• Prefer clear, text-based PDFs over scanned images when possible.
• Split huge binders into logical files if answers feel incomplete.
• Keep filenames meaningful for @ mentions and Library browsing.
• Re-upload when a source document is substantially revised.

Related

• Upload Documents
• Search and Retrieval
• Common Issues