Skip to main content

Agent Settings

Overview​

The Agent Settings screen allows administrators to configure and manage all key properties of an Agent, including API keys, embedding models, connectors, tags, and organization information.

Agent Settings management interface showing key metadata, embedding selection, JSON-LD option, and API keys

Embeddings​

Choose the appropriate text embedding model based on your language and cost needs:

  • ADA V2 (text-embedding-ada-002) – The original embedding model. Consider upgrading for better performance and cost-efficiency.
  • Text Embedding 3 Small (Recommended) – Offers improved accuracy at lower cost. Best for English-only Agents.
  • Text Embedding 3 Large – Recommended for multilingual Agents; higher cost, but better support for non-English content.

Embedding model selection interface

CMS Connectors​

CMS Connectors must first be enabled in the Organization Settings. Once enabled, you can activate connectors at the Agent level for integrations with supported content platforms like OneDrive or CMS systems.

API Keys​

API keys enable access to the Agent via web components (search/bot controls) or via developer integrations (REST API or WebSocket).

Creating an API Key​

  1. Go to the API Keys section.
  2. Click Create API Key.
  3. Enter a descriptive name (e.g., Web, Mobile, Analytics).
  4. Click Save. Your new key appears in the list with status and creation details.

Create API Key form

List of API Keys showing active keys with names, creation date, and action menu

Managing API Keys​

To edit, archive, or delete an API key:

  1. Click the action menu (three dots) next to the key.
  2. Select Archive to deactivate or Edit to rename.

Action menu for API Key management

💡 Tip: Use meaningful key names to easily filter usage logs and manage access across applications.

Agent Organization Information​

This section contains AI-generated context about the organization the Agent was created for. You can update this content at any time to improve the quality of responses and categorization during knowledge base creation.

Click Edit to make changes.

Replace Words​

The Replace Words feature lets you substitute certain terms in real-time across AI responses. This is useful for standardizing terminology or personalizing output.

  • Not case-sensitive
  • Example: Replace ai1toz with ai12z

Replace Words interface showing substitution list and input fields

Categories​

Manage content categorization for your Agent here:
→ Go to Category Management

reRank (default is disabled)​

Use reRank only if you're using an expensive language model and your results contain fewer than 10 vector matches. This can reduce token usage and improve performance when using a small number of vector documents.

JSON-LD Support​

ai12z automatically detects and ingests JSON-LD from web pages. This format offers highly structured data ideal for semantic search and knowledge extraction.

You can also choose to append JSON-LD content during ingestion by enabling this setting.

Auto Tag Content​

When enabled, this feature will:

  • Automatically applies tags to ingested content
  • Generates content titles and descriptions
  • Calls GPT-4o-mini for each file (incurs a small cost)

⚠️ This can affect billing depending on content volume.

Connectors​

This is where you select the CMS connectors

Tags​

Tags can be used to include or exclude certain content during search. Useful for segmenting data in use cases like eCommerce.

Danger Zone: Delete Agent​

⚠️ Warning: Deleting an Agent will permanently remove all data associated with it. This action cannot be undone.