How to add knowledge to Serpio

By the end of this you will have a source in your Knowledge Base marked Ready, with settings that say how Serpio may use it. The form takes two or three minutes; processing usually finishes a few minutes later. You need a site in your workspace, and adding knowledge does not spend article credits.

Before you start

• A site in your workspace. Knowledge Base in the sidebar opens the one for the selected site. See How to add a new site.
• For a document: a PDF, DOCX, Markdown, or TXT file under 20 MiB that contains real text. Scanned PDFs are rejected; Serpio does not run OCR.
• For a URL: a public https:// address that loads without a login. Private hosts and custom ports are refused.
• A decision on whether its facts may appear in articles or it is background only.

Steps

1. 1

   Open the Knowledge Base

   Click Knowledge Base in the sidebar. You see the Knowledge base heading, the collapsed Website crawler panel, and the source list with the filters All, Documents, Added URLs, and Website.

2. 2

   Press Add knowledge

   The Add knowledge modal opens on the Upload documents tab, with Add URL beside it.

3. 3

   Add files, if uploading a document

   Click Choose files or drop them here, or drag files onto it. Each file appears as a row reading Ready to upload. A file that fails validation shows why in red, such as Choose a file no larger than 20 MiB. Remove takes a row out.

4. 4

   Or paste an address, if adding a URL

   Press Add URL and fill URL with the full address. Optionally add a Title · optional and a note under How should Serpio use this? · optional about what the page is for.

5. 5

   Set Use this as and Permission for writing

   Pick Brand knowledge for material about your own product, or Reference evidence, the default, for material that backs up claims. Then pick Facts allowed in articles or Background context only; for context only the form explains Use for understanding and context. Its private facts and quotations are not approved for article copy.

6. 6

   Scope and schedule

   Tick Content focuses if the source belongs to specific focuses; the note says none selected means all. For a URL, set Check URL for changes to Manually, Daily, or Weekly, the default. Optionally set Effective from and Review by; the form warns that after the review date the source is excluded from writing until you update it.

7. 7

   Submit

   Press Upload documents or Add URL. Document rows move through Uploading with a percentage, Upload received; queueing processing…, then Queued for processing. Follow progress in your knowledge base. For a URL the modal closes at once. Then press Close.

8. 8

   Watch processing and confirm Ready

   A strip above the list reads Knowledge queued for processing…, then Extracting source content…, then Indexing knowledge… It becomes available when indexing succeeds., then Processing complete. Your knowledge base is up to date. The new row then shows your settings and status, for example Reference evidence · Facts allowed · Ready, with Version 1 beneath. Open it to check the Preview.

What happens next

Once the version is Ready, every article you start can retrieve passages from the source. Idea collections and existing articles are not changed, and their quality reports stay as they were. On a new article, the Knowledge used panel lists the passages taken from your source, with the version number.

URL sources on Daily or Weekly are re-checked on that cadence and re-indexed only when the page changed. Documents change only when you replace them. See How to keep your Knowledge Base current.

If something looks off

• A PDF fails with a message that it appears to be image-only. It has no text layer. Export it again from the original, or run OCR in another tool, and upload the text version.
• The URL is rejected with Enter a full http:// or https:// URL. or Private network URLs are not supported. Include the scheme and use a public host with no credentials or custom port.
• The strip ends with an error instead of Processing complete. Open the source; the drawer shows it under Latest processing issue. Fix the cause, then press Reprocess original or Check URL now.
• The row reads Usage settings pending. Open it, expand Source settings, confirm the two settings, and press Save settings.
• A new article did not use a Ready source. Retrieval keeps only passages that fit the brief, and skips sources marked Background context only.

Go deeper

• Knowledge Base overview for what the Knowledge Base is and is not.
• Adding documents for formats, limits, and the no-OCR rule.
• Usage settings and permissions for what each setting does at write time.
• Versions, statuses, and source actions for every label you will see.
• Serpio's site index is not Google's index for the website crawler.