Uploading documents
Upload a PDF, a spreadsheet, or a Word document from your dashboard and Contextick turns it into text your AI can bring back later. No AI is involved in the upload itself, so the length of the document costs you nothing.
What uploading does
An upload converts your file to text and saves that text to your knowledge layer. It runs in three steps, and only the first two need you:
- Reading. Your file is converted to plain text. A large file takes about half a minute, and the page has to stay open while it runs.
- Saving. The text is stored right away. From this moment it is in your list and you can open and read it.
- Processing. In the background, the saved text is prepared so an AI can find it by meaning. This is the part that finishes later.
Only the text is kept. The file itself passes through the converter on its way back to your browser and is not stored.
What you can upload
PDF, DOCX, XLSX, CSV, ODT, and ODS files, up to 15 MB each. You can upload 10 files an hour.
A PDF has to contain real text. A scanned document is a picture of a page, so there is nothing to extract from it and those are not supported.
If a document needs more space than you have left, the upload screen tells you before it saves.
For text you already have on screen, paste it in instead — no file needed, and it is stored exactly as you wrote it.
Why one file becomes many items
A long document is saved as several numbered items rather than one enormous one, because a single item holds a limited amount of text. They appear in your list as Report (1/12), Report (2/12), and so on, each carrying the summary you wrote once.
This is not a workaround. Recall works on meaning, and a smaller self-contained passage matches a question more precisely than a whole book would — so your AI gets the part that answers you rather than the entire document.
The parts stay linked to each other, which is what makes "delete the whole file" and "retry the whole file" possible from any one of them.
While it is processing
Your document is saved and readable straight away, but an AI can only find it once background processing finishes. That usually takes a few minutes, and up to 24 hours when the queue is busy. Your list shows where each item stands:
| What you see | What it means |
|---|---|
| Processing | Saved and readable. Not findable by an AI yet. |
| Delayed | The same thing, still going after 24 hours. |
| Not searchable | Processing failed for this item — see below. |
| No badge | Finished. Your AI can find it. |
You do not have to keep the page open for any of this — once saving is done, the rest is ours. When everything you uploaded has finished, we email you a summary of what is ready and what is not.
When something goes wrong
An item marked Not searchable is saved and readable; only its background processing failed, so no AI will find it. Because a long document is stored as several items, this often affects just part of one.
Open any item from that document and you get two buttons:
- Try again. Runs the failed parts again. The text is already stored, so there is nothing to upload a second time and nothing to convert again.
- Delete all. Removes every item that came from that file in one go, including the ones that succeeded.
If other parts of the same document are still processing, Try again appears once they finish — there is nothing to decide until the document has settled.
Deleting and uploading again is not the repair. That re-converts the file and re-runs every part, including the ones that already worked. Trying again redoes only what failed.
More
What your AI can do with an uploaded document — find it, read it, delete it — is covered in the MCP docs, along with the limits that apply to everything you save. Common questions are answered in the FAQ, plans are on pricing, and if something is unclear or broken, get in touch — a real person reads every message.