...
Node Type: This is the type of node and must be selected according to the purpose of the node (giving a message to the customer, asking the customer for information, querying an external web service, etc.). The available node types are (find the detailed description of each node type below):
Ask
Say
Capture: By selecting this node type, the following additional fields will be displayed in the form:
Expected: Type here a regular expression to validate the response sent by the customer.
List: By selecting this node type, the following additional fields will be displayed in the form:
List: Specify here the list of values to be shown to the customer for selection. In case the tickets come from a previous node of Web Service type, use the nomenclature {{webhook.response}}.
Fallback: When selecting this node type, the following additional fields will be displayed in the form:
Bot: Select here the bot to which the conversation will be moved in case of exceeding the maximum attempts.
Bot Node: Select here the bot node to which the conversation will be moved in case the maximum attempts are exceeded.
Maximum Attempts: Specify here the maximum number of attempts that the customer will have to retry the previous step to this node.
Scheduled Node: When selecting this node type, the following additional fields will be displayed in the form:
Start Date: Specifies here the start date of the period/range during which this node will be applied.
End Date: Specify here the end date of the period/range during which this node will be applied.
Validator: When selecting this node type, the following additional fields will be displayed in the form:
Condition Type: Specifies here whether the validation rules will be applied with the conditional and (or - all of the rules comply) or with the conditional or (or - some of the rules comply).
Validation Rules: Add one or more validation rules, specifying a variable, an operator, and a value.
Web Service: When selecting this node type, the following additional fields will be displayed in the form:
Request Type: Select one of the available HTTP request types (GET, HEAD, POST, PUT, DELETE, CONNECT, OPTONS, TRACE, PATCH).
URL/Endpoint/IP: Specify here the URL/endpoint/IP address to access the webservice. For example, https://mywebservice.mybusiness.com/getCustomerInfo/4224563
Headers: Specify here the headers that should be added to the HTTP request to consume the web service. For example, { "Authorization": "AccessKey GoCsdWE5rR2x7oXTQ4cPn4fTb5R"}.
Body: Specify here the body of the HTTP request. Generally, this body should have a JSON structure. For example, {"first name": "Jane", "last name": "Doe}.
Timeout: Specify here the maximum time Sagicc will have to wait when consuming the web service. If this time is exceeded, the request will be considered failed.
Dialog Flow
Go to
Bot: Select here the bot to which the conversation will be moved.
Bot Node: Select here the bot node to which the conversation will be moved.
Gen AI: When selecting this type of node, the following additional fields will appear in the form:
Prompt: Write the prompt here that will be used to give provide instructions to the AI agent, ensuring that you include the communication tone, context, expectations, and any other relevant details that to effectively guide its behavior.
Enable RAG technology: Select this option if you want to enable RAG (Retrieval-Augmented Generation) technology in the Gen AI node. This technology will allow the node to access articles from Sagicc’s Knowledge Base to generate more accurate and contextually relevant responses.
Knowledge Base Categories for RAG: If you enable RAG technology for the node, select in this field the categories from Sagicc's Knowledge Base Keep in mind that the success of implementing a bot powered by Generative AI depends largely on the quality of the prompt provided. Find more information about Generative AI here: https://aws.amazon.com/es/what-is/generative-ai/. Learn more about Prompt Engineering here: https://aws.amazon.com/es/what-is/prompt-engineering/.
Enable RAG Technology: Select this option if you want to enable RAG (Retrieval-Augmented Generation) technology in the Gen AI node. This technology allows the node to access Sagicc's Knowledge Base articles to generate more precise and contextually relevant responses. Find more information about RAG technology here: https://aws.amazon.com/es/what-is/retrieval-augmented-generation/.
Knowledge Base Categories for RAG: If RAG technology is enabled for the node, select the Sagicc Knowledge Base categories in this field that contain articles with information relevant to the AI Agentagent.
| Info |
|---|
Remember Keep in mind that the success of implementing a bot powered by Generative Artificial Intelligence largely depends on the quality of the prompt provided. It is essential that it contains clear and well-defined instructions. Make sure to be specific in the directions included in the prompt, covering details about customization, the bot's expected behavior, and any constraints that should be considered. The more precise and detailed the prompt is, the more effective the bot's interaction with users will be.
|
use of this type of node involves additional costs due to the use of Artificial Intelligence models. These costs are determined by the number of input tokens and output tokens calculated for each interaction with the AI agent. Below, we explain the key concepts related to this process.
Example: If a client sends a query that consumes 50 input tokens, and the model generates a response of 60 output tokens, the total tokens processed will be 110 tokens. Costs are based on this total number of tokens, which varies according to the length and complexity of both the input and the output. It’s important to consider these details, as using more tokens increases the cost of the service. |
Use template: Activate this option if the message you want to send in the node corresponds to a previously created template in Sagicc. By activating it, you will be able to select the desired template.
...