Skip to content
AskVault Docs
Try Free →

File uploads to the knowledge base

Written by Joseph Russo, Principal Architect, Networks and Tech, AskVault · Reviewed by Allison Chan
Last updated: · 4 min read

Supported file types by plan

File-type support scales with plan, so a higher tier unlocks additional formats: each tier includes everything below it:

  • Free. PDF, TXT, Markdown, CSV.
  • Starter. + DOCX / DOC (Microsoft Word).
  • Growth. + HTML, and images (PNG, JPEG, WebP, GIF): text extracted via OCR.
  • Business. + XLSX (Excel), PPTX (PowerPoint), and video (MP4, MOV, WebM, MKV, M4V). Audio transcribed automatically.
  • Enterprise. Every supported format.

Uploading a file your plan doesn't support returns an error naming the format and pointing at an upgrade. There's no need to convert PPTX to PDF or XLSX to CSV on Business and above: both are parsed natively.

Format notes:

  • PDF. Text-based PDFs index fully; scanned (image-only) PDFs need OCR pre-processing on plans below Growth (Growth+ can instead upload the scan as an image and let AskVault's built-in OCR handle it).
  • DOCX. Text in the body of the document, tables included, is indexed. Keeping each table in place with its header row, and labelling every piece with its section heading, is rolling out (see below).
  • CSV / XLSX. Text is extracted and split into chunks. Row-aware chunking, which keeps each row whole with its column names, is rolling out. Rolling out
  • Images. Extracted via multimodal OCR: useful for whiteboard photos, screenshots, architecture diagrams.
  • Video. Audio transcribed via multimodal transcription; the transcript is what gets indexed.

Uploading a file

Upload files from the dashboard:

  1. Open Knowledge > Add Source > Document Upload.
  2. Drag and drop one or more files into the upload area.
  3. Click Upload. The file joins the indexing queue straight away.

There is no public API for uploading files today.

What happens during indexing

The pipeline:

  1. File received. Stored encrypted at rest immediately.
  2. Parser runs. A format-specific extractor pulls the text out of the file.
  3. Chunker splits content into small, meaning-sized passages.
  4. Embedder generates vectors for each chunk.
  5. Vectors stored in the workspace's index.
  6. Document marked ready. Available for retrieval.

How long it takes depends on the file's length and layout. Watch the status column on the Knowledge page; the file is ready when it shows as indexed.

Watch progress under Knowledge > [document] which shows "Queued > Indexing > Ready" with a percentage.

Per-file size limit

The per-file cap follows your plan:

  • Free. 5 MB per file.
  • Starter. 15 MB per file. Starter+
  • Growth. 40 MB per file. Growth+
  • Business. 100 MB per file. Business+
  • Enterprise. 500 MB per file.

For files at or above your plan's limit:

  • Split into smaller files. PDF tools can split by chapter.
  • Compress images in the source document before uploading.
  • Convert to text-only. A scanned PDF can be 10x larger than the OCR-extracted text equivalent.
  • Upgrade the plan for a higher per-file cap.

If a file exceeds your plan's limit at upload, the dashboard rejects it and shows the limit.

Knowledge content cap

Your plan has one knowledge-content cap. It is the same number the pricing page shows:

  • Free. 5 MB. Roughly 50 to 100 pages of typical text.
  • Starter. 15 MB. Roughly 150 to 300 pages. Starter+
  • Growth. 40 MB. Roughly 400 to 800 pages. Growth+
  • Business. 100 MB. Roughly 1,000 to 2,000 pages. Business+
  • Enterprise. 2 GB.

Today that cap measures your crawled website content. Uploaded files are not counted against it.

Rolling out, not yet switched on: uploaded files will count toward the same cap, alongside your website content. There will be no separate storage limit. Once this is on:

  • An uploaded file counts at its file size, from the moment it is accepted.
  • An upload that would take you past the cap is refused before it is stored. The message shows how much of the cap is already used and links to upgrading.
  • Deleting a file gives its space back straight away.
  • Content imported from connectors (Notion, GitHub, Confluence and others) is still not counted.

Check current usage under Knowledge > Storage Usage.

File-type indexing details

How each format is handled:

PDF (text-based). Text extracted page by page, in order. Headings are not detected, and tables come through as lines of text without their column layout.

PDF (scanned/image). Indexing produces empty results without OCR. Pre-process via Adobe Acrobat (File > OCR), Tesseract (free CLI), or an online tool. Then re-upload.

DOCX. Text in the body of the document, tables included, is indexed. Headers, footers, footnotes, text boxes, content controls and tables nested inside another table are not, and bullet and numbering marks are not kept. Today each table is read after the body text, one line per row, with the column names only on the table's first line, and heading styles are not used as section labels. Rolling out (not yet switched on): tables stay where they appear, with their first row as the header, and each piece carries the heading it sits under. Word files uploaded after the switch get the new layout; delete and re-upload an existing one to update it. Accept or reject all tracked changes before uploading: text inserted with Track Changes and not yet accepted is not indexed. Comments are not indexed.

TXT. Indexed line-by-line. No structural inference; chunks split at paragraph breaks.

Markdown. Headings (H1-H6) used for chunk boundaries. Code blocks indexed as code; fenced language preserved.

CSV / XLSX. Useful for FAQ tables, product catalogues and other structured data. Today the text is split by length, so a long row can be cut in two; row-aware chunking that keeps every row whole with its column names is rolling out. XLSX requires Business or above.

PPTX. Slide text indexed per slide. Requires Business or above.

HTML. Indexed like a saved web page. Requires Growth or above.

Images (PNG, JPEG, WebP, GIF). Text extracted via multimodal OCR. Requires Growth or above.

Video (MP4, MOV, WebM, MKV, M4V). Audio transcribed via multimodal transcription; the transcript is indexed. Requires Business or above.

Audience after upload

Every uploaded file starts with Audience public. To keep one out of retrieval on public channels, set it to internal after indexing:

  1. Open Knowledge and open the document's settings.
  2. Pick an Audience: public or internal. There are no other values.

There is no bulk audience control, and you can't choose an audience while uploading.

Replacing or updating a file

There is no in-place replace. When the source document changes, delete the old document, then upload the new version. The new upload gets a new document ID, so answers cite the new file from then on. Delete first: until you do, both versions are indexed and either can be cited.

Citations in bot responses

When the bot answers from an uploaded file, the source shows the file name, so the customer can see where the answer came from. Page numbers are not attached to PDF answers today.

OCR pre-processing for scanned PDFs

Many older PDFs are scanned images, not text. Without OCR, AskVault indexes nothing useful.

Three OCR options:

  • Adobe Acrobat Pro. File > Recognize Text.
  • Tesseract (free CLI). tesseract input.pdf output -l eng pdf. Open-source, runs locally.
  • Online tools like ilovepdf.com or smallpdf.com.

After OCR, upload the OCR'd PDF (now text-searchable) to AskVault.

Bulk upload

For uploading many files:

Dashboard. Drag and drop several files at once; each one is queued and indexed on its own.

Zip archives. Today, AskVault doesn't auto-extract zip files. Unzip first, then bulk-upload.

Planned features (on the roadmap)

Documented for accuracy:

  • Zip archive auto-extract. Today, unzip manually and upload the contents. Auto-extract planned.
  • Table-extraction improvement. Today, table parsing is best-effort. Improved table-to-CSV conversion planned for complex layouts.

Limits

  • Per-file size. Plan-dependent: 5 MB (Free) up to 500 MB (Enterprise). See per-file size limit.
  • File types. Plan-dependent: see supported file types by plan. No file-count cap; only total indexed content counts against your plan's storage cap.
  • Total content per workspace. Plan-dependent (5 to 100 MB on paid tiers below Enterprise; custom on Enterprise).

Common pitfalls

PDF uploaded but no answers reference it. Scanned PDF without OCR. Run OCR and re-upload.

Indexing stuck at "Queued". Other files are ahead of it in the queue. Give it a few minutes; if it does not move, contact support with the file name.

Workspace hit storage cap. Free plan covers 5 MB. Either delete old documents or upgrade.

Table content garbled in answers. Complex PDF tables lose structure during extraction. Upload the source CSV alongside for better table queries.

File rejected as "unsupported format". Either the extension isn't supported at all (check the supported types), or it's gated to a higher plan than yours: e.g. DOCX needs Starter+, XLSX/PPTX need Business+.

FAQ

Can I upload files larger than my plan's limit?

Not per single file. Split into multiple files, each under your plan's per-file cap, or upgrade for a higher cap.

Do I need to OCR PDFs myself?

For scanned PDFs on Free or Starter, yes. On Growth and above, upload the scan as an image instead (PNG/JPEG/etc.) and AskVault's built-in OCR extracts the text automatically.

How long does indexing take?

It depends on the file's length and layout. The Knowledge page shows when each file is indexed.

Can I bulk upload via API?

No. There is no public API for uploading files today. Drag several files into the upload area at once instead.

What happens to my files if I delete the workspace?

The workspace and its indexed files are removed. This can't be undone, so export anything you need before deleting.

Was this page helpful?