Skip to main content
Populating your knowledge base is the first step toward building an AI agent that delivers accurate, source-grounded responses. Beenos Solutions gives you three ways to bring content in — file uploads, web URL crawling, and direct text entry — so you can work with whatever format your content already lives in. Once added, each source is automatically processed and indexed, usually within a couple of minutes.

​
Add a New Source

Upload local files directly from your computer. Beenos Solutions parses and indexes the full text content of each file you provide.
1

Open your knowledge base

Navigate to the Knowledge Base section from the left sidebar and select the knowledge base you want to add content to. If you haven’t created one yet, click New Knowledge Base and give it a name.
2

Click Add Source

On the knowledge base detail page, click the Add Source button in the top-right corner, then select Upload a Document from the dropdown.
3

Select or drop your file

Drag and drop a file onto the upload area, or click Browse Files to open your system file picker. You can upload one file at a time. The maximum file size per upload is 50 MB on the Free plan and 500 MB on Pro and Enterprise plans.
4

Confirm and upload

Review the detected file name and format, then click Upload. Beenos Solutions will begin parsing and indexing the file immediately. The source will appear in your sources list with a Processing status badge while indexing is underway.

​
Supported Formats

Beenos Solutions can parse the following file types when you upload a document. Unsupported formats are rejected at upload with an error message.
Full text extraction is supported for both native PDFs and scanned PDFs with embedded OCR text. Scanned PDFs without a text layer are not currently supported — convert them using an OCR tool before uploading.
Content is extracted from the body text, headings, and tables. Comments, tracked changes, and embedded macros are ignored during parsing.
Uploaded as-is with no formatting transformation. Encoding must be UTF-8. Files in other encodings should be converted before uploading.
Markdown and MDX files are parsed and indexed as plain text. Frontmatter fields (such as title and description) are extracted and stored as source metadata.
Each row is indexed as a separate chunk. The first row is treated as a header and used to provide column context for each subsequent row. Large CSVs with thousands of rows may take longer to index than other file types.
Give every source a clear, specific name before saving it. Descriptive names like Q4 2024 Product Changelog or EU Privacy Policy make it much easier to identify sources in search results, manage them later, and understand which content your agents are drawing from.
Most sources finish indexing within 1–2 minutes of being added. Large files, deep recursive URL crawls, or CSVs with many rows may take up to 10 minutes. Your agents cannot query a source until its status badge changes from Processing to Indexed.