AI Builder: Smarter Data Linking Across AI Actions & Create Task

Overview

We've shipped major improvements to how AI Builder links data into AI actions and the Create Task action, making generated workflows significantly more accurate and reliable.

What's New

Smarter Data Linking Across AI Actions

AI Builder now correctly passes the actual variable or data into action fields instead of inserting redundant prompts or instructions. Previously, AI Builder would inject a full prompt into an action's input field (e.g., "Summarize the following text into bullet points…") rather than simply passing the variable holding the value. Since these actions already know what to do, the extra instructions were redundant and sometimes caused unexpected output.

Now, AI Builder cleanly passes just the relevant variable (e.g., {{message.body}}, into each action's input field, letting the action perform as designed.

Actions improved:

  • AI Summarize — Passes the target variable directly into the input field instead of wrapping it in summarization instructions.
  • AI Translate — Passes the source text variable cleanly, no longer adds translation prompts on top.
  • AI Intent Detection — Correctly links the content variable for sentiment or intent analysis.
  • AI Decision Maker — Passes the standard value for the content on which the decision needs to be made. Also resolves an issue where branch names were being left empty by AI Builder.

Screenshot 2026-04-02 at 4

Create Task Action Improvement

The task description field now contains an actual, human-readable task description for the assignee, not a prompt instructing the system to create a task.

Before: Description would read something like "Create a follow-up call task for the new lead."

After: Description reads something like "Follow up with {{contact.name}} to discuss their inquiry and schedule a discovery call."

This makes the generated task immediately actionable for the person it's assigned to.

Why This Matters

These improvements make AI Builder-generated workflows production-ready out of the box — less manual cleanup, fewer broken automations, and a much smoother experience from prompt to live workflow.

AI Builder: Smarter Data Linking Across AI Actions & Create Task

Overview

We've shipped major improvements to how AI Builder links data into AI actions and the Create Task action, making generated workflows significantly more accurate and reliable.

What's New

Smarter Data Linking Across AI Actions

AI Builder now correctly passes the actual variable or data into action fields instead of inserting redundant prompts or instructions. Previously, AI Builder would inject a full prompt into an action's input field (e.g., "Summarize the following text into bullet points…") rather than simply passing the variable holding the value. Since these actions already know what to do, the extra instructions were redundant and sometimes caused unexpected output.

Now, AI Builder cleanly passes just the relevant variable (e.g., {{message.body}}, into each action's input field, letting the action perform as designed.

Actions improved:

  • AI Summarize — Passes the target variable directly into the input field instead of wrapping it in summarization instructions.
  • AI Translate — Passes the source text variable cleanly, no longer adds translation prompts on top.
  • AI Intent Detection — Correctly links the content variable for sentiment or intent analysis.
  • AI Decision Maker — Passes the standard value for the content on which the decision needs to be made. Also resolves an issue where branch names were being left empty by AI Builder.

Screenshot 2026-04-02 at 4

Create Task Action Improvement

The task description field now contains an actual, human-readable task description for the assignee, not a prompt instructing the system to create a task.

Before: Description would read something like "Create a follow-up call task for the new lead."

After: Description reads something like "Follow up with {{contact.name}} to discuss their inquiry and schedule a discovery call."

This makes the generated task immediately actionable for the person it's assigned to.

Why This Matters

These improvements make AI Builder-generated workflows production-ready out of the box — less manual cleanup, fewer broken automations, and a much smoother experience from prompt to live workflow.

AI Builder: Smarter Data Linking Across AI Actions & Create Task

Overview

We've shipped major improvements to how AI Builder links data into AI actions and the Create Task action, making generated workflows significantly more accurate and reliable.

What's New

Smarter Data Linking Across AI Actions

AI Builder now correctly passes the actual variable or data into action fields instead of inserting redundant prompts or instructions. Previously, AI Builder would inject a full prompt into an action's input field (e.g., "Summarize the following text into bullet points…") rather than simply passing the variable holding the value. Since these actions already know what to do, the extra instructions were redundant and sometimes caused unexpected output.

Now, AI Builder cleanly passes just the relevant variable (e.g., {{message.body}}, into each action's input field, letting the action perform as designed.

Actions improved:

  • AI Summarize — Passes the target variable directly into the input field instead of wrapping it in summarization instructions.
  • AI Translate — Passes the source text variable cleanly, no longer adds translation prompts on top.
  • AI Intent Detection — Correctly links the content variable for sentiment or intent analysis.
  • AI Decision Maker — Passes the standard value for the content on which the decision needs to be made. Also resolves an issue where branch names were being left empty by AI Builder.

Screenshot 2026-04-02 at 4

Create Task Action Improvement

The task description field now contains an actual, human-readable task description for the assignee, not a prompt instructing the system to create a task.

Before: Description would read something like "Create a follow-up call task for the new lead."

After: Description reads something like "Follow up with {{contact.name}} to discuss their inquiry and schedule a discovery call."

This makes the generated task immediately actionable for the person it's assigned to.

Why This Matters

These improvements make AI Builder-generated workflows production-ready out of the box — less manual cleanup, fewer broken automations, and a much smoother experience from prompt to live workflow.

AI Builder: Smarter Data Linking Across AI Actions & Create Task

Overview

We've shipped major improvements to how AI Builder links data into AI actions and the Create Task action, making generated workflows significantly more accurate and reliable.

What's New

Smarter Data Linking Across AI Actions

AI Builder now correctly passes the actual variable or data into action fields instead of inserting redundant prompts or instructions. Previously, AI Builder would inject a full prompt into an action's input field (e.g., "Summarize the following text into bullet points…") rather than simply passing the variable holding the value. Since these actions already know what to do, the extra instructions were redundant and sometimes caused unexpected output.

Now, AI Builder cleanly passes just the relevant variable (e.g., {{message.body}}, into each action's input field, letting the action perform as designed.

Actions improved:

  • AI Summarize — Passes the target variable directly into the input field instead of wrapping it in summarization instructions.
  • AI Translate — Passes the source text variable cleanly, no longer adds translation prompts on top.
  • AI Intent Detection — Correctly links the content variable for sentiment or intent analysis.
  • AI Decision Maker — Passes the standard value for the content on which the decision needs to be made. Also resolves an issue where branch names were being left empty by AI Builder.

Screenshot 2026-04-02 at 4

Create Task Action Improvement

The task description field now contains an actual, human-readable task description for the assignee, not a prompt instructing the system to create a task.

Before: Description would read something like "Create a follow-up call task for the new lead."

After: Description reads something like "Follow up with {{contact.name}} to discuss their inquiry and schedule a discovery call."

This makes the generated task immediately actionable for the person it's assigned to.

Why This Matters

These improvements make AI Builder-generated workflows production-ready out of the box — less manual cleanup, fewer broken automations, and a much smoother experience from prompt to live workflow.

Clarifying Agent in AI Builder

The AI Builder now gathers missing context before generating or editing workflows, producing significantly better first outputs and reducing the need for follow-up revisions.

What's New

Previously, when a user provided a vague prompt like "follow up with the customer," the AI Builder would make assumptions and generate a full workflow — often requiring multiple rounds of revision. The same applied to edit requests: vague instructions led to changes that didn't always match user intent. Each revision cycle passed the entire workflow in context, increasing time to value.

The new Clarifying Agent detects gaps in the user's request and surfaces up to three focused questions before generation or editing begins. By collecting the right inputs upfront, the AI Builder delivers a more accurate first output for both new builds and edits, with fewer revisions and lower cost per interaction.

ezgif

When Does It Activate?

The Clarifying Agent only intervenes when specific, high-impact details are missing:

  • Trigger not specified — The user hasn't defined what initiates the workflow (e.g., form submission, appointment booked, tag added).
  • Channel not provided — The user hasn't indicated the communication channel (e.g., SMS, email, WhatsApp).
  • Timing not defined — The user hasn't specified when an action should occur (e.g., immediately, after one hour, after one day).
  • Unsupported platform referenced — If the user mentions a channel that isn't supported, the agent flags it and asks which supported channel to use instead.

How It Works

Open the AI Builder and describe the workflow you want to build or the edit you want to make.

If key details are missing, the Clarifying Agent surfaces targeted questions before the AI proceeds.

Select from the provided options or type a custom response.

Skip any question to let the AI decide on your behalf.

The AI Builder generates or edits the workflow with all additional context incorporated.

Questions are presented in a clean multiple-choice format. When multiple questions are surfaced, navigation arrows allow users to move between them. Every question is skippable.

Why This Matters

The quality of any AI-generated output depends on the context it receives. The Clarifying Agent ensures the AI Builder has the right inputs before it begins, leading to better first outputs, fewer post-generation revisions, and a faster path from idea to live workflow.