AI tools

AI tools help you delegate parts of scenario configuration and decision-making to AI systems.

Let AI Decide (one-click placeholders)

In most fields where AI Agent or MCP can provide values, you will see a Let AI Decide button. It inserts the correct operator (fromAIAgent or fromMCP) into the field and auto-generates a suitable parameter name/description.

fromAIAgent()

Use fromAIAgent() inside any editable field of a node that is connected to an AI Agent. This marks the field as an argument the agent must provide during execution. The agent sees these placeholders as tool parameters and fills them automatically.

For full context, see: AI Agent Node.

Format:

{{fromAIAgent("parameter_name"; "description")}}

Example:

{{fromAIAgent("Email Body"; "Include an email body as either plain text or HTML. If HTML, make sure to set the \"Body Type\" prop to html")}}

fromMCP

fromMCP is used in MCP workflows to mark inputs and fields that an MCP client should decide and fill. In many places you can also use Let AI Decide to insert the correct placeholder automatically.

For setup details, see: MCP Nodes.

MCP Trigger input parameters

In MCP Trigger → Tool configuration → Input parameters, add parameters and set:

• Key: parameter name (e.g., email, message)
• Type: fromMCP
• Description: what the AI client should pass

MCP Response

By default, MCP returns the output of the last node. MCP Response lets you return exactly what the client needs — and you can use Let AI Decide to make the response MCP-fillable.

askAI()

The askAI() operator sends a request to the built-in AI and returns a text response.

In addition to text, the request can use existing variables or global variables or output parameters of previous nodes, enclosed in symbols according to the pattern "+Variable/Data+".

Below are some examples of using askAI().

Text Generation

A user request can be a text prompt, such as asking to generate an invitation for an event as the value of the variable Val:

1. Add the Trigger on Run once and setvariables nodes to the scenario.

2. Add the variable Val and set its value to {{askAI("Generate a short invitation text for an event")}}.

3. Run the scenario once and review the node execution results to verify the presence of the new variable.

User Feedback Monitoring

A request can involve identifying the tone or sentiment of incoming text. The text can be the output parameter of a previous node, such as an email or Telegram chat message. For simplicity, generate the text directly within the scenario by adding the following nodes:

1. Trigger on Run once node to trigger the scenario with the Run once button.

2. setvariables node to generate the Text variable, containing the product review text.

3. Webhook response node to return a response upon successful execution of the scenario. In the Body field of the Webhook response node, add the AI operator with a request using the variable from the setvariables node: {{askAI("Determine if the text \"" + _.Text + "\" is a negative review")}}

The result of this scenario is an AI response:

Yes, the text "The packaging is damaged and the courier was late" can be considered a negative review.

Text Classification

A request can involve determining if the incoming text is a question. Using the AI operator in routes allows the scenario to follow different branches based on the AI's response.

For simplicity, generate the text directly within the scenario by adding the following nodes:

1. Trigger on Run once node to trigger the scenario with the Run once button.

2. setvariables node to generate the Value variable containing the text for classification.

3. Question route with the condition {{askAI("The text contains \"" + $2.Value + "\" is there a question? If so, return one word \"true\", otherwise return one word \"false\"") = "true"}}.

4. Webhook response node for the Question route with the response The text contains a question upon scenario execution.

5. Not a question route with the condition {{askAI("The text contains \"" + $2.Value + "\" is there a question? If not, return one word \"true\", otherwise return one word \"false\"") = "true"}}.

6. Webhook response node for the Not a question route with the response The text does not contain a question upon scenario execution.

The result of the scenario depends on the text in the Value variable:

• If the variable contains a question, such as What is the deadline for completing the task?, the scenario's result is The text contains a question.

• If the variable contains a statement, such as Documentation is an important part of learning, the scenario's result is The text does not contain a question.