## Knowledge

### Overview

By leveraging knowledge, you can empower your agents to understand your business, playbooks, interfaces, and much more. For every skill you teach an agent, you have the flexibility to choose the datasets you want them to be familiar with.

### Datasets

Datasets empower you to manage and organize your knowledge effectively. Build knowledge bases that your agents can easily access to enhance their capabilities. To create a new dataset, simply click "+ Dataset…" in the sidebar. Then, upload files or webpages to train your agents on the information.

### Webpages

Much of your business information is already available on your website or in cloud-based documents. You can train your agents by providing public URLs for them to scrape. With just one click, they will retrieve all the text and images from the website. To keep the content current, simply click "Refresh" under the three-dot menu. Remember, your webpages must be publicly accessible, as the system cannot access documents stored behind an authorization screen.

### Files

You likely already have an existing knowledge base that your human agents rely on. You can effortlessly upload your files using drag-and-drop or the "Upload File" button, enabling your Ai agents to become knowledgeable about them.

We support a variety of formats, including:

**Base types:** pdf.

**Documents and presentations:** 602, abw, cgm, cwk, doc, docx, docm, dot, dotm, hwp, key, lwp, mw, mcw, pages, pbd, ppt, pptm, pptx, pot, potm, potx, rtf, sda, sdd, sdp, sdw, sgl, sti, sxi, sxw, stw, sxg, txt, uof, uop, uot, vor, wpd, wps, xml, zabw, epub, Images, jpg, jpeg, png, gif, bmp, svg, tiff, webp, web, htm, html.

**Spreadsheets:** xlsx, xls, xlsm, xlsb, xlw, csv, dif, sylk, slk, prn, numbers, et, ods, fods, uos1, uos2, dbf, wk1, wk2, wk3, wk4, wks, 123, wq1, wq2, wb1, wb2, wb3, qpw, xlr, eth, tsv.

**Audio (** limited to 20mb **):** mp3, mp4, mpeg, mpga, m4a, wav, webm.

For instance, you can upload your PDF playbook or screenshots of an app to equip your agents with an understanding of the content. Lemni is completely multi-modal. Which means that it understands both images, audio, and text.