Validate Message
The Validate Message endpoint uses Artificial Intelligence (AI) to validate a message using the specified criteria.
Endpoint POST
https://api.ttgtoolbox.com/v1/validate/message
Request body
This endpoint uses the POST method and requires a JSON object as the request body.
Required parameters
The request body requires the following parameters.
messagestringrequired
The message to validate. The message can be from any source, including non-standard formatting characters, such as paging pipes (|).
criteriastringrequired
The validation criteria. The criteria provided dictate the specific aspects of the message that the AI validation API will assess for accuracy, relevance, or compliance.
Criteria suggestions & examples
Check Fields: Do the message fields contain specific or required information?
Example: "The caller is Luke Skywalker and he's arrived at Alderaan."
Grammar and Spelling: Ensuring the message is grammatically correct and free of spelling errors.
Example: "Please check if the message adheres to proper grammar and spelling."
Content Relevance: Verifying that the message content aligns with the intended purpose or topic.
Example: "Verify if the message content is relevant to ..."
Sensitive Information Detection: Identifying sensitive information, such as personal identifiers like SSNs or credit card numbers.
Example: "Check for the presence of sensitive information, such as Social Security Numbers or credit card details."
Language Tone: Assessing the tone of the message to ensure it matches the desired level of formality or politeness.
Example: "Evaluate the language tone to confirm it aligns with ..."
Optional parameters
modelstring
The default LLM for message validation is OpenAI's GPT-3.5 Turbo gpt-3.5-turbo, which performs effectively for basic messages. Nevertheless, you have the flexibility to explore other LLM options by specifying the model parameter.
| Model | model value |
|---|---|
| OpenAI GPT-3.5 Turbo | gpt-3.5-turbo default |
| OpenAI GPT-4 Turbo | gpt-4-turbo |
| OpenAI GPT-4 | gpt-4 |
| Cohere Command | cohere.command-text-v14 |
Additional models may be added in future releases. Costs vary depending the chosen LLM.
Example request body
{
"criteria": "Caller's Name, Company, Account, Phone Number, Product, Outage",
"message": "Message Ticket: \nCaller:| Jim Hall |\nCompany:| Acme |\nAccount:| 1029 |\nPhone:| 555-719-3559 |\nProduct: eFax\nOutage: Yes\nMsg:| Not getting notice of faxes anymore. . . Not getting failed or successful faxes. |"
}Example request body with optional parameters
{
"model": "gpt-4",
"criteria": "The caller is Luke Skywalker and he's arrived at Alderaan.",
"message": "Message Ticket: Caller:| Luke Skywalker |Company:|| Account:| 11111 | Phone:| 800-555-1212 | \nMsg:|. he's caught in a tractor beam and being pulled into a space station.|"
}Valid message response
{
"valid": true,
"message": "All requirements for the message have been met.",
"credits": 1
}Invalid message response
{
"valid": false,
"message": "The requirements were not met. The caller, Luke Skywalker, was supposed to have arrived at Alderaan. However, the message states that he's caught in a tractor beam and being pulled into a space station.",
"credits": 1
}Notes
Be advised
The Message Validation API uses a Large Language Model (LLM) to generate the validation results. Repeated calls to the API using the exact text will produce different responses. Also, note that other models behave differently depending on the specific criteria and message. It may take some experimenting to determine what works best for you and your particular use case.