Initial Setup

Before using CTERA Content Services, you have to set up the following:

• At least one LLM
• At least one collector
• System settings
• Sources that you can query
• Datasets for the sources
• An expert to enable querying the sources

Adding LLMs

A Large Language Model (LLM) is a type of AI deep learning model trained on massive datasets to understand, generate, and summarize text or other content. Based on transformer architectures, LLMs act as next-word predictors.

To add an LLM:

1. Sign in as an administrator, using the following address https://<ip>/admin/login
   
   The administration dashboard is displayed.
2. Select LLMs in the navigation pane.
   
3. Click New LLM.
   
4. Under LLM Details, enter the following:
   Name – A name for the LLM.
   Provider – Select an LLM provider from the list.
   Model – Select a model from the list.
5. Under Endpoint, enter the API Key.
6. Click Save Changes.

Adding a Collector

A collector is designed to gather and organize the massive amounts of data required to train or operate AI models (like LLMs), which require high-quality datasets to learn.

To add a Colllector:

1. Select Collectors in the navigation pane.
   
2. Click New Collector.
   
3. Enter the following details in the General tab:
   Name – A name to identify the collector.
   URI – The URI to access the collector.
   Default Collector – Slide on to make this collector the default collector.
   Collector Status – Select whether you want the status to be Enabled or Disabled
4. Click Connect in the URI field to connect to the collector and retrieve file type mappings.
   
5. For each file type, select the Reader if you don't want the default.
6. Click Save Changes.

Managing System Settings

1. Select Settings in the navigation pane.
   The System tab for Settings is displayed.
   
2. Enter the following details:
   Ingest Workers – The number of ingestor workers for scaling.
   Embedding Mode – The embedding mode to use from the drop-down list. The port must be open. See Inbound Ports.
   Embedding Model – The embedding model to use from the drop-down list.
   Embedding Service URL – The URL for the embedding mode.
   Sparse Embedding Model – The sparse embedding model to use from the drop-down list.
   FastEmbedService URL – The fast embed service URL. The port must be open. See Inbound Ports.
   Default Model – The LLM that will be used as the default, from the LLMs defined to CTERA Content Services.
   Classification Model – The LLM model to use for data classification. If enabled, select the model from the drop-down list of LLMs defined to CTERA Content Services.
   Summarize Model – The LLM to use to create summaries. .
   Vision-Language Model (VLM) – The model to use for the VLM. If enabled, select the model from the drop-down list of LLMs defined to CTERA Content Services.
   Personalized Chat Memory – Save chat prompts across the user conversations.

Adding Sources

To add a source:

1. Select Sources in the navigation pane.
   
2. Click New Source.
   
3. Enter the source details:
   Name – A name to identify the source.
   Type – A type of source from the drop-down list.
   Host – The URL to access the data source.
   If the Type is Web you also specify the number of directories to crawl or 0 for an unlimited number of directories.
4. Enter the authentication details. The required details depend on the Type.
5. Enter whether you want the source status to be Enabled or Disabled and click Save Changes.

Adding Datasets

To add a dataset:

1. Select Datasets in the navigation pane.
   
2. Click New Dataset.
   
3. Enter the dataset details:
   Name – A name to identify the dataset.
   Source – A source from the drop-down list. The source list includes the sources defined in Sources.
   Capabilities – The search and tagging capabilities to apply to this dataset.
   Collector – The type of collector to use.
   Owner – The dataset owner.
   Collaborators – The users who can collaborate on the dataset.
4. Add content to the dataset. The selected content section is dependent on the source. For example, when the source is a CTERA Portal, the content includes cloud drive folders that belong to a specific owner or without any owner.
   1. Click to add a folder.
      
   2. Click Users to expand the users and then select a user and from the list of cloud folders, select a folder.
      The folder content is displayed.
      
   3. Click Add Folder
5. Enter the schema details and then the indexing actions required.
6. Click Save Changes.

Adding an Expert

Experts are task-focused AI agents, that act as virtual employees to respond to specific types of enquires using your organization’s curated content. For example, a marketing expert might have access to all the company marketing material and competitive information, while a developer expert might have access to all the CTERA SDK and its examples.

To create an expert:

1. Sign in as an administrator, using the following address https://<ip>/admin/login
   
   The administration dashboard is displayed.
2. Click Experts in the navigation pane.
   
3. Click New Expert.
   
4. Enter the expert details:
   Expert Name – A name to identify the expert.
   Expert Subtitle – A short description of the expert.
5. Upload a jpeg or png file to use as the expert avatar.
6. Select the expert owner from the drop-down list and collaborators for this expert, from the drop-down list.
   If you specify any collaborators, only these users will have access to the expert.
7. Slide Set as Public Expert on if you want this expert to be available publicly.
8. Slide Set as Default Expert on if you want this expert to be the default expert for all users.
9. Slide Expert Enabled on to make the expert active and available to users.
10. Select the LLM to use for this expert in the LLM Model field.
11. Select the datasets to use for this expert in the Datasets field.
12. Slide the tools on under Expert Tools that will be used with this expert, for example Semantic Search and Image Generator.
13. Optionally , select an MCP Server from the MCP Servers list.
14. Click Save

To test the expert:

1. In a separate browser tab, enter https://<ip>/ where is the IP address of CTERA Data Intelligence to open the end user view.
   The end user dashboard is displayed.
   
2. Write a message and verify you receive an answer.

Or:

1. Click the expert.
   
2. In the Expert Settings panel, click Test Expert.