Add AI Knowledge Documents
Table of Contents
Your AI Knowledge Documents form the foundation of your AI features. They are used by your AI Agents to generate accurate and helpful responses, as well as any other useful content.
To improve the quality of these responses, start by reviewing and refining your documents. They should be as complete, clear, precise, and well-organized as possible.
Please note
- Do not reference any documents that contain confidential company information.
- Only reference documents for which you own the intellectual property or for which you have a usage agreement.
Your knowledge documents

Go to the “Knowledge Documents” tab in the “Settings / AI Alcmeon” section to view all your saved documents. From there, you can edit existing ones or create new documents by clicking the “New Document” button.
Knowledge Documents and GAIA
Please note that knowledge documents referenced in Alcmeon can only be used by your AI Agents using GAIA, Alcmeon's generative AI engine. If you are using your own AI engine or one provided by a partner, it will need to rely on its own method for referencing and accessing documents.
Name
By clicking on it, you can edit your document’s settings.
Active / inactive status
If a document is set to “inactive”, it will no longer be used by any AI Agents it's linked to.
Document type
This indicates the source of the content—e.g. a public website, text excerpt, or PDF document.
Integration status
Some documents, like websites, may take time to be fully indexed by Alcmeon before they become fully usable by your AI Agents.
This status allows you to track the progress of the document indexing process. It can take several values:
- Pending (gray): Indexing has not yet started
- Started (yellow): Document is currently being indexed
- Completed (green): Indexing is finished, and the document is now ready to be used by your AI Agents
- Error (red): Indexing failed: We recommend reviewing the document for possible errors and restarting the indexing process.
Last synchronization date
Indicates the most recent date the document was indexed. This is particularly useful for website-type documents, which can be configured for regular, periodic indexing.
AI Agent Usage
A single knowledge document can be used by one or more AI Agents. This section shows the list of AI Agents currently linked to the document.
DELETE
This button allows you to delete the document. It will no longer be indexed by Alcmeon or available to your AI Agents. Please note: This action is irreversible. All content associated with the document will be permanently removed from Alcmeon's infrastructure.
Add or edit a knowledge document

First, configure the following settings:
- Document Name: For internal use only within Alcmeon
- Active: Only documents marked as “active” can be used by your AI Agents
- Used by AI Agents (optional): If you’ve already created AI Agents, you can select which ones should have access to this document
- Document Type: Choose whether the content is a pdf, public website, or text excerpt.
PDF document

Note: “Image PDFs” cannot be used. Only “Text PDFs” or “PDF/A” file types are supported.
A PDF is indexed only once, after it has been uploaded.
If you edit a PDF knowledge document, you can replace the original PDF with a new version. In this case, the previous version will be deleted from your knowledge documents in Alcmeon.
Public website type document

URL
In the “URL” field, enter the address of the root page of the website you want to add to your knowledge base. This could be the homepage of your website, or the main page of an FAQ or Help section, for example.
When you create a website knowledge document, Alcmeon will crawl it by navigating through it and following links on the same domain. For instance, if you create a knowledge document for https://help.alcmeon.ai, and it contains links to https://alcmeon.ai or https://wikipedia.org, these external links will be ignored. However, links from the same domain but with different directories (e.g. https://alcmeon.ai/service/ and https://alcmeon.ai/company/) will be indexed. If you want to index pages from multiple domains, please create a seperate knowledge document for each domain.
Adding a website-type knowledge document may involve indexing multiple pages, which can take some time.
The process may also be blocked by anti-bot protection or encounter broken links (pages that no longer exist). If no pages are indexed, the status will show as “Failed” in red in the list page. To resolve this, make sure the site doesn't have a robots.txt file that prevents indexing, and that the pages are accessible.
Update frequency
You can choose how often Alcmeon will re-index your website content in case of changes.
By default, it is recommended to leave the setting at “once a month” or “once a week” to avoid overloading the document indexing process.
Depth of followed links
You can set the depth of links to be followed ranging from 0 and 5, with te default value set to 3.
A depth of 0 means only the page specified in the URL field will be indexed. A depth of 1 will crawl and index links found on the first page, stopping there. A depth of 2 will also crawl links on the pages indexed at depth 1, and so on.
Force content refresh now
This button appears when you are editing an already created document. It allows you to trigger Alcmeon to re-index your website immediately.
Text Extract type document

A text snippet is the quickest to add new information to your knowledge base. It is processed only once, either when it is created or updated. Only the most recent version of the text will be stored in the knowledge base.
The text field cannot exceed 5000 characters.
Additionally, a code injection check is performed to ensure that html, javascript or other code is excluded. If such code is detected, it will not be saved.