```html

Artificial Intelligence: Integration Configuration with OpenAI

Terms of Use

The use of the software implies the sending and processing of information in generative artificial intelligence systems maintained by third parties, which may be located in jurisdictions abroad. By using the software, the user declares their full knowledge and agreement with this fact, and must observe, at a minimum, the following restrictions:

The sending of sensitive or confidential information, in any format, should be avoided, as the information will be processed and stored on third-party servers, which are beyond the control or supervision of LUMIS, and may also be used to improve artificial intelligence systems. If you choose to send sensitive or confidential information, especially those listed by the General Data Protection Law, the user declares to have the powers and legitimacy to do so, being aware of the risks and acting within legal limits, taking full responsibility for this conduct;
There is no guarantee of originality or factual accuracy in the responses or results. The results may contain excerpts from quotes of third-party works, or violate the legislation in effect in the user's jurisdiction. It is up to the user to verify whether the information obtained is in accordance with current legislation, especially copyright laws, and free from plagiarism, as well as whether they are correct in relation to facts, dates, and other factual characteristics;
LUMIS does not assume any responsibility for the results of the use, or for the legitimate or illegitimate, lawful or unlawful use of the results obtained from the use of the software. The user, hereby, exempts LUMIS from any joint or subsidiary liability in this regard, under any circumstances;
The use of the systems for unethical purposes, such as the preparation of academic or professional work whose authorship must be exclusive and mandatory for the user, is not allowed;
The user declares to be aware and in agreement with the terms of use of the ChatGPT system, available at https://openai.com/policies.

By using the software, the user declares, on their own behalf and on behalf of the organization they represent, if applicable, that they have the necessary powers to make this use, agree to these terms, and exempt LUMIS from any responsibility for their use, including with regard to damages to third parties, except in cases of willful misconduct by LUMIS.

Integration configuration details page with OpenAI. After filling in the fields properly and saving, the AI resources are made available through the portal.

Features

It is an administration interface

Screens

Basic Information
- API Key (string): Key used to connect with the OpenAI API. If empty, the artificial intelligence functionalities will be disabled.
- AI Model for embedding calculation (string, required field): Defines the embedding model used for the request. Ex: text-embedding-ada-002
- Block size for embedding calculation (characters) (integer, required field): Maximum size of fragments generated, in number of characters, when calculating embeddings of a text.
Content Indexing
- Enable embedding calculation in indexing for search (boolean): Indicates whether the default indexing for search should calculate embeddings of indexed texts. Calculating embeddings implies sending the corresponding text to be processed by OpenAI
- Enable embedding calculation in indexing for private content searches (boolean): If disabled, only the indexing of content that is permitted to be viewed by all users will have its embeddings calculated. If enabled, the indexing of any content will have its embeddings calculated. Calculating embeddings implies sending the corresponding text to be processed by OpenAI.
Content Assistant
- Enable content assistant (boolean): Defines whether the OpenAI content assistant API will be enabled or not.
- AI Model used in content assistant (string, required field): Artificial intelligence model that will be used by the content assistant. Ex: gpt-4o-mini
Virtual Assistant
- Maximum size (in number of characters) of the knowledge fragment (integer, required field): Defines the maximum number of characters that a knowledge fragment can have to use with the virtual assistant. Example: 2048.
- Model for chat completion (string, required field): Artificial intelligence model that will be used by the virtual assistant. Ex: gpt-4o-mini
- Model for text analysis (string)
- Maximum number of knowledge items searched per interaction (integer, required field): Defines the maximum number of knowledge items that the AI searches for in each interaction. Default value: 5.
- Maximum number of interactions in processed context (integer, required field): Specifies the maximum number of interactions that the Virtual Assistant will perform in the context. Default value: 3.
Customer Experience AI Assistant
- Enable Customer Experience AI Assistant (boolean, required field, default value: true): Defines whether the customer experience assistant will be enabled in the portal or not. By default, this option is already enabled.
- Model for chat completion (string, default value: gpt-4-turbo-preview): Artificial intelligence model that will be used by the customer experience assistant. Ex: gpt-4o
- Maximum number of interactions in processed context (integer, default value: 5): Defines the number of interactions that will be conducted in the processed context.
- Variation intensity (integer, default value: 30): Value from 0 to 100 that indicates the intensity of variation of the result generated by the artificial intelligence. Lower values generate responses understood by the artificial intelligence as more common or expected, and higher values generate more varied responses, providing a more creative effect but with the possibility of lower accuracy.
- Additional instructions for AI (text): Field that defines custom instructions that can be entered to customize the behavior of the customer experience assistant. Example:
  - To ensure that reports are organized by the channel path, write in the field: When presenting data grouped by channel, prefer to group by the channel path instead of by the channel name.

Permissioning

Edit content of service instance: Allows managing all data of this service instance with publisher profile.

Actions

commit

```