Documents
Documents are the files a project knows about — manuals, datasheets, schematics — and this page covers how to add them and keep them healthy.
Expert user / administratorDocuments live inside a project. They are the files Edmund reads when it answers a question about that machine — typically PDF manuals, datasheets, schematics and spare-parts catalogues. Once a file is uploaded, Edmund extracts its text, indexes it, and can search inside it and cite the exact page. A document Edmund can’t read is a document it can’t answer from, so it’s worth checking that each one is in good shape.
Upload a document
Uploading is done from inside the project, so open the project first.
Open the project you want to add knowledge to.
Go to Documents.
Choose Upload documents at the top right and pick the file or files. Most projects use PDF manuals, datasheets, schematics and spare-parts catalogues.
Wait for the row to show Indexed in the Status column. Until then, Edmund hasn’t finished reading the file.
Reading the status column
The documents list is a table with columns for Name, file extension, Status, Type and the date it was created. The Status column tells you whether Edmund can actually use the file:
- Indexed — Edmund extracted the text and indexed it. The document is searchable and can be cited in answers.
- No text — no text could be extracted. This is almost always a scanned or image-only PDF with no selectable text layer, so Edmund can’t read it yet. The file is still listed, but nothing inside it reaches your answers.
The Type column is filled in automatically. Edmund classifies each document into a category such as Component Document, Mechanical Drawing or Spare Parts Catalog, and marks it Unknown when it can’t tell.
Fix a “No text” document
A document that shows No text contributes nothing to answers, even though it appears in the list. If a manual seems to be ignored, check its status here first.
The fix is to give Edmund a version of the file that has real, selectable text, then upload that:
- Re-export the file from its original source with a proper text layer, if you have the original.
- Otherwise, run OCR on the scan to add a text layer, then save the result.
- Upload the corrected file. The new row should reach Indexed. Delete the old No text row once it’s replaced.
Find and manage documents
Once a project has a long list, use the controls above the table to narrow it down. You can filter by file extension, by Status and by Type, and search by name. Filtering by No text is a quick way to see everything that still needs fixing.
Each row has a menu, and its only action is Delete. Use it to remove a superseded or broken file.
Delete is the only per-row action, and it is permanent. There is no archive or undo — if you delete a document by mistake, you’ll need to upload it again.
What makes a good source document
Edmund works from text, so the best sources are text-based files: the actual manual, datasheet or schematic export rather than a photo of a page or a screenshot. A clean PDF with a real text layer indexes well and gives precise, citable answers. A scan of the same pages may land as No text and add nothing. When you have a choice, upload the original document over a captured image of it.
