Create an AI profile - Fluid Topics

This web service allows users with the ADMIN role to create an AI profile.

It is necessary to provide an Ft-Calling-App value when using Fluid Topics web services. See Fluid Topics calling app.

For more information about AI profiles, see AI.

Method	Endpoint
`POST`	`/api/admin/ai/profiles`

Request example for API key authentication

{
  "profileType": "CHATBOT",
  "name": "My Chatbot",
  "profileId": "my_chatbot",
  "modelEndpoint": {
    "provider": "openai",
    "providerLabel": "OpenAI",
    "host": "",
    "modelOrDeploymentId": "gpt-5"
  },
  "customPrompt": "Insert your prompt here",
  "numSemanticSearchResults": 5,
  "authenticationMethod": {
    "type": "api_key",
    "label": "API key",
    "authenticationParameters": [
      {
        "label": "API key",
        "key": "api_key",
        "isSecret": true,
        "isOptional": false,
        "value": "XXX"
      }
    ]
  }
}

Request example for username/password authentication

{
  "profileType": "TRANSLATION",
  "name": "Phrase profile",
  "profileId": "phrase_profile",
  "modelEndpoint": {
    "provider": "phrase",
    "providerLabel": "Phrase",
    "modelOrDeploymentId": ""
  },
  "customPrompt": "",
  "authenticationMethod": {
    "type": "username_password",
    "label": "Username and password",
    "authenticationParameters": [
      {
        "label": "Username",
        "key": "username",
        "isSecret": false,
        "isOptional": false,
        "value": "Username"
      },
      {
        "label": "Password",
        "key": "password",
        "isSecret": true,
        "isOptional": false,
        "value": "Password"
      }
    ]
  }
}

Field	Type	Required?	Description
`name`	String	Yes	The name of the profile.
`profileId`	String	Yes	The profile ID of the profile. It can only include alphanumeric characters, hyphens (`-`), and underscores (`_`).
`profileType`	String	Yes	The type of the profile. Can be `AGENTIC_CHATBOT`, `COMPLETION`, `CHATBOT`, `GENERIC_AGENT`, or `TRANSLATION`.
`modelEndpoint`	Object	Yes	An object containing the model details.
`provider`	String	Yes	For `CHATBOT`, and `COMPLETION` profiles, can only be one of the following values: `anthropic`, `azureopenai`, `fluidtopics`, `googleai`, `myllm`, `openai`. For `TRANSLATION` profiles, can be `deepl_pro`, `intento`, or `phrase`.
`providerLabel`	String	No	Enter the name of the provider: `anthropic`: `Anthropic` `azureopenai`: `Azure OpenAI` `deepl_pro`: `DeepL Pro` `fluidtopics`: `Fluid Topics` `googleai`: `Google AI` `intento`: `Intento` `myllm`: `MyLLM` `openai`: `OpenAI` `phrase`: `Phrase`
`host`	String	Yes (only with an `azureopenai` or `myllm` provider)	The resource name. For example: `my-resource.openai.azure.com`.
`modelOrDeploymentId`	String	Yes	The LLM model or translation engine. It is possible to enter any value, as long as it matches a valid model from the selected LLM provider. A list of models and engines is available in the AI menu. It is not exhaustive. Enter an empty value for DeepL Pro.
`authenticationMethod`	Object	Yes	Contain the details for the authentication to the provider.
`type`	String	Yes	The type of authentication. Possible values are `username_password` or `api_key`.
`label`	String	Yes	The label for the type of authentication.
`authenticationParameters`	Object	Yes	The list of parameters to connect to the provider.
`label`	String	Yes	The label for the authentication parameter.
`key`	String	Yes	The key for the authentication parameter.
`value`	String	Yes	The value for the authentication parameter.
`isSecret`	Boolean	Yes	Whether the parameter is secret or not.
`isOptional`	Boolean	Yes	Whether the parameter is optional or not.
`customPrompt`	String	Yes	Only with `AGENTIC_CHATBOT`, `CHATBOT`, or `COMPLETION` profiles. The LLM prompt. The prompt can contain variables. Place variables in single curly brackets. Users can set the values of variables when using the AI method in Custom components, or the Query a large language model (LLM) web service.
`numSemanticSearchResults`	Number	No	Only with `AGENTIC_CHATBOT`, or `CHATBOT` profiles. The number of semantic search results provided to the large language model (LLM) when calling the Retrieval-augmented generation web service. The value can range from `1` to `20`. By default, the value is `5`. A higher number of semantic search results can increase the cost of requests, and decrease the speed and quality of answers. For most use cases, use the default value of `5`. The Chatbot component calls the Retrieval-augmented generation web service.
`numSemanticSearchResultsAfterRerank`	Number	No	Only with `AGENTIC_CHATBOT`, or `CHATBOT` profiles. The number of semantic search results provided to the large language model (LLM) after reranking. The value must be smaller or equal to the value of the `numSemanticSearchResults` field.
`ragPrompt`	String	No	Only with `CHATBOT` profiles. A Retrieval-augmented generation (RAG) prompt. Asks the user question to the LLM. It must have the `{relevant_documents}` variable, and the `{chatbot_instructions}` variable for Chatbot profiles. `{relevant_documents}` dynamically contains the relevant documents (gotten through the Clustered semantic search web service), and `{chatbot_instructions}` is the `customPrompt` value.
`ragHistoryPrompt`	String	No	Only with `CHATBOT` profiles. A Retrieval-augmented generation (RAG) history prompt. Asks the LLM to transform the user question and the conversation history into a standalone question asked to the LLM.
`translationAdvancedSettings`	String	No	Only with `TRANSLATION` profiles. Parameters for the translation engine. Accepts JSON as a string.
`translationDestinationLanguages`	Array	No	Only with `TRANSLATION` profiles. A list of target languages available to end users in On-demand translation components that use this profile. The expected format depends on the translation provider. For example, it may be `fr` or `fr-FR` for French.
`relevantMetadata`	String	No	Only with `AGENTIC_CHATBOT` and `GENERIC_AGENT` profiles. `relevantMetadata` allow the agent to use metadata filters for their queries. `relevantMetadata` contains a stringified JSON array of metadata objects, each containing a `key` identifier, an array of `values`, and a `description`. Proper description on how and when to use metadata filters is essential. If the agentic chatbot runs a query with filters and gets no response, it runs the query again.
`promoteRules`	String	No	Only with `AGENTIC_CHATBOT` and `GENERIC_AGENT` profiles. `promoteRules` uses metadata to indicate which variant of a document is the most relevant in a document cluster. For example `"version": "latest"`. `promoteRules` contains a stringified JSON array of metadata objects, each containing a `key` identifier, and an array of `values`. If the variant does not exist, the agentic chatbot ignores the rule. Rules are applied in order.
`agentPrompt`	String	Yes	Only with `AGENTIC_CHATBOT` and `GENERIC_AGENT` profiles. The `agentPrompt` defines specific instructions and the conversation scope for the bot. For optimal results, add specific instructions that are relevant to your context. Below are a few examples: "If the user is asking about product _X, search for release notes X before answering." "_Make sure you have properly identified the product the user is referring to." "Suggest to create a support ticket and provide a link to _support.example.com if the user found a critical bug." "_If the relevant documentation contains a topic named "Prerequisites" or "Before you begin", provide its content to the user."
`agentTools`	Array	No	Only with `GENERIC_AGENT` profiles. `agentTools` defines tools that the agent has access to. The only available value is `semsearch`.
`responseJson`	String	Yes	Only with `GENERIC_AGENT` profiles. The stringified JSON schema representing the format of the response. `responseJson` cannot contain the top-level field `title`.

profileId and name values must be different from existing AI profiles.
For username/password authentication, the authenticationParameters object requires a list of string values for the username and one for the password, as shown in the example.

See Chatbot architecture for a better understanding of the use of Chatbot prompts and reranking.

Response body

{
  "id": "7ed9433f-af15-40df-8f32-15beddce37df",
  "name": "Phrase profile"
}

Field	Type	Description
`id`	String	The unique identifier for the profile.
`name`	String	The name of the profile as set in the request.

Return code	Description
`200 OK`	The profile was created successfully.
`401 UNAUTHORIZED`	The authorization header is absent or invalid.

For a comprehensive list of all possible return codes, see Return codes.

Create an AI profile - Fluid Topics - Latest

Fluid Topics API Reference Guide