Knowledge Base overview: what Serpio knows about your brand

Serpio writes from material it can check. The Knowledge Base is where you give it that material: your own documents, public pages you point it at, and the pages it finds on your site. Open it from the sidebar under Knowledge Base. The page heading reads Knowledge base and the subtitle names the brand it belongs to. Every brand has its own Knowledge Base, and nothing in one brand's Knowledge Base is visible to another.

Three kinds of sources

Every row on the page is a source. Sources come in three kinds, and the filter chips above the list (All, Documents, Added URLs, Website) show one kind at a time.

Documents

Files you upload: PDF, DOCX, Markdown, or TXT, up to 20 MiB each. Serpio extracts the text, splits it into passages, and indexes them. The file itself stays in Serpio's private storage. See Adding documents.

Added URLs

Public pages you add by address. Serpio fetches the page, keeps the readable text, and can re-check it daily, weekly, or only when you ask. See Adding URLs and refresh policies.

Website

Pages the website crawler found on your own domain. These arrive automatically after you add a site and are recrawled weekly. The crawler lives in the Website crawler panel at the top of the page. See Serpio's site index is not Google's index.

Use the Search all knowledge… box to find a source by title, description, or topic. Results are paged 25 at a time, with Previous and Next at the bottom. Refresh list reloads the list without changing anything.

What the page tells you

The banner under the heading is the short version of this guide. It reads: "Add documents or URLs, choose how they can inform writing, and keep their facts current. Files stay private. New knowledge informs future work; existing published articles are not changed automatically." Each row below it shows:

• The title, or the filename for a document, or the address for a URL.
• A usage label and a status, for example Reference evidence · Facts allowed · Ready. The first two parts are the source's settings; the last is its current state.
• When it was last checked or indexed, and the active version number, for example Version 3.
• Any warning in amber, such as a legacy excerpt that needs a refresh.

Click a row to open its drawer. The drawer holds the action buttons, a Preview of the indexed text, the Source settings, and the Version history. Versions, statuses, and source actions explains every label.

How sources inform articles

Idea discovery and the Knowledge Base are separate. The content engine prepares ideas from public research and the app stores them as weekly collections. Only after you choose a topic and start an article does Serpio search the Knowledge Base with that article's brief, pull the best passages, and hand them to the writer as evidence.

The search is hybrid: meaning-based matching plus keyword matching, so a distinctive product name is found even when its wording does not resemble the query. Every hit is then re-checked against the live source settings. A source that is paused, past its review date, not yet effective, scoped to a different focus, or marked context only is dropped before the writer sees it. Retired versions never reach a new draft.

On a finished article, the Knowledge used panel lists the passages that were used, with the version number of the source each came from. Public web sources are listed separately under Public sources.

Privacy

Uploaded files stay private in Serpio's storage and index, and Serpio denies model providers permission to collect or train on your data. Filenames and storage paths are stripped from the passages the writer sees, so a private document is never named in article copy. If you want a fact to carry a public citation, the source needs a public URL; see Usage settings and permissions.

What the Knowledge Base does not do

• It does not read scanned PDFs. Image-only PDFs are rejected because there is no text to extract, and Serpio does not run OCR.
• It does not sync with Google Drive, Notion, or any other cloud document tool. You upload files or add URLs.
• It does not rewrite drafted or published articles when knowledge changes.
• It is not Google's index. The Website crawler builds Serpio's own private index for writing and internal links.
• It does not publish your files anywhere. Sources feed the writer; they are never exported with an article.
• It does not guarantee that a given fact will appear in a given article. Retrieval picks the passages that fit the brief.

Get started in three steps

1. 1

   Add a source

   Press Add knowledge. Upload a file under Upload documents, or switch to Add URL and paste a public page address. Product docs, pricing pages, and positioning notes are good first sources.

2. 2

   Set how it may be used

   In the same modal, choose Use this as and Permission for writing. The defaults are Reference evidence and Facts allowed in articles. Pick Background context only for anything that should inform tone but never be quoted.

3. 3

   Wait for Ready

   A status strip shows Knowledge queued for processing…, then extraction and indexing, then Processing complete. Your knowledge base is up to date. The row reads Ready, and the next article you start can use it.