Skip to main content

Usage and Differences Between “Dynamically fill with AI” and “Custom value” in AI Agent Tool Inputs

Learn the difference between “Dynamically fill with AI” and “Custom value” when configuring tool inputs, and understand how to choose the right method for building accurate and reliable AI Agent logic.

Updated yesterday

When configuring an AI Agent’s tools, each tool input (Inputs) must specify a value source. Yeeflow provides two commonly used fill options:

  • Dynamically fill with AI

  • Custom value

Understanding how these two options work—and when to use each one—will help you build smarter, more reliable AI Agent tool logic.


1. Dynamically fill with AI: Input values generated automatically by AI

When Dynamically fill with AI is selected, the AI automatically generates the input value based on the current context.
The AI considers:

  • The user’s natural-language input

  • The AI Agent’s Persona & Prompt

  • The Tool’s Description

  • The Input parameter’s Display name and Description (especially important)

Using these references, the AI infers and constructs a suitable value for the tool input.

When to use this option

  • The input value depends on natural-language expressions

  • The value needs to be extracted, analyzed, or generated by AI

  • The expected value cannot be predetermined (e.g., dynamic filters)

Example

A user says: “Show me all red products that are low in stock.”

AI may automatically generate:

Color == "Red" and Inventory < 10

Important notes when using Dynamically fill with AI

To ensure the AI generates correct values, you must provide clear guidance in at least one of the following:

  • Describe the field’s purpose in the Prompt

  • Explain expected logic in the Tool Description

  • Define the field format and meaning in the Input Description (most recommended)

The clearer the description, the more accurate the AI inference.


2.Custom value: A fixed value that does not rely on AI generation

When using Custom value, you manually provide the input value, and the AI does not attempt to infer or construct anything.

When to use this option

  • The value is fixed and does not change based on user input

  • The filter or expression structure must remain constant

  • You do not want the AI to generate or guess the expression format

  • The tool requires strictly formatted expressions (such as Data filter)

Example

Fields = "ProductName, Inventory" Sorts = "CreatedOn desc" Page size = 20

Note: Custom value can also reference AI Agent input variables

Although Custom value is often used for fixed values, it can also reference predefined Input variables from the AI Agent when building expressions such as Data filter.

In the expression editor, you can select:

👉 AI Agent → Input variables

For example:

  • Input.ProductName

  • Input.Threshold

  • Input.Category

These variables receive their values dynamically from user input, forms, workflows, or API requests.

Example expression

ProductName == Input.ProductName and Inventory < Input.Threshold

The Data filter remains a Custom value, but its parameters become dynamic.


3. Key differences between the two fill methods

Item

Dynamically fill with AI

Custom value

Value generated by AI

✔ Yes

✘ No (fixed or variable reference)

Depends on natural language

✔ Yes

✘ No

Flexibility

High

Medium

Stability

Medium (depends on AI inference)

High (format controlled)

Suitable for strict filter format

✘ Not recommended

✔ Highly recommended

Can reference Input variables

Rarely used

✔ Fully supported


4. Best practices

✔ If the filter structure is fixed → Use Custom value (+ Input variables)

Ensures stability and predictable logic.

✔ If the value depends on natural-language context → Use Dynamically fill with AI

Ensures more natural and flexible user interaction.

✔ Always provide Input Descriptions

Clear descriptions help the AI understand how to generate values.


Summary

  • Dynamically fill with AI
    Best for scenarios where AI must infer or extract values from natural language dynamically.

  • Custom value
    Best for fixed-format, predictable, and controllable logic, and can also reference AI Agent input variables to achieve a powerful combination of fixed structure + dynamic parameters.

By choosing the appropriate fill method, you can create AI Agents that are both intelligent and reliable.

Did this answer your question?