Skip to main content

​
Summary

Drafts are one of the most important control points in TruAgents. They are where campaign intent, agent behavior, approval requirements, and sending infrastructure all meet before something becomes a live communication.

​
Who this is for

  • Operators reviewing outbound quality
  • Campaign managers validating messaging before send
  • Anyone trying to understand how a generated message becomes a real communication

​
The key distinction

  • a draft is pre-send and reviewable
  • a communication is the actual sent or received event
That distinction matters because problems can happen at either layer. A draft can look wrong before send, or a communication can look wrong after delivery because of surrounding run, provider, or thread context.

​
What the current product suggests

The draft model appears to include:
  • status transitions
  • approval and rejection paths
  • retry behavior
  • channel-aware differences
  • links to contact, campaign, and run context
  • infrastructure-related blockers such as missing delivery setup

​
The draft lifecycle in plain English

Simple version:
  1. a campaign or workflow generates a draft
  2. the draft enters a status
  3. a human or system decision may be required
  4. the draft is approved, rejected, retried, or fails
  5. if it moves forward successfully, it becomes part of live communications history

​
Why approval is more than a button

Approval is operational because it sits between generation and sending. That means approval decisions can protect the team from:
  • low-quality content
  • wrong context
  • premature launch
  • infrastructure problems that would only be discovered later

​
Draft state thinking

The current product clearly implies at least three broad classes of states:
  • drafts still being processed
  • drafts waiting for user action
  • drafts in final states such as success, rejection, or failure
That distinction matters because the right response changes by state.

​
What can make a draft look wrong

Sometimes the draft content is the problem. Sometimes the real issue is:
  • missing email or phone setup
  • campaign configuration problems
  • contact-data problems
  • channel-specific delivery constraints
So the team should avoid assuming every bad draft is “an AI quality issue.”

​
Good review habits

  • read the draft with campaign intent in mind
  • check whether the draft is blocked by setup before editing content mentally
  • use history before retrying
  • separate content objections from infrastructure or workflow objections
  • remember that a draft is a pre-send decision point, not just a preview pane

​
Common misunderstandings to avoid

  • Drafts are not just a draft folder.
  • Approval is not just a cosmetic step.
  • A failed or stuck draft may indicate an infrastructure issue, not only a content issue.
  • A successful draft state is not the same thing as understanding the eventual communication outcome.
  • when to require human approval
  • when retry is appropriate
  • when the real fix belongs in setup instead of content
  • when a recurring draft issue points back to agent or campaign design