Skip to main content
Back to NLP Topics Amend Entity lets users change entity values during task execution. Example: a user books a flight and then changes the destination or travel date mid-conversation.
Amend Entity is supported only for English dialog task conversations. Scope: entity value replacement only (not partial modifications, deletions, or additions).

How It Works

  1. Amend Identification — The platform’s built-in NLP detects the user’s intent to amend.
  2. Amending Entities — Three flows based on utterance:
    • User refers to entity name: prompted for new value. (If entity appears multiple times, the most recently populated one is amended.)
    • User provides entity value: entity is updated. (If value matches multiple entities, user selects which to amend.)
    • User refers to both entity name and value: entity is updated directly.
  3. Post-Amendment Behavior — Three options:
    • Re-execute dialog from the amended node (evaluating its connections).
    • Resume from the node where the amend was identified.
    • Resume from a specified node.

Configuration

Amend behavior is set at two levels. Task-level settings override app-level.

App Level

  1. Go to Natural Language > NLU Config > Amend Entity.
  2. Enable Allow amend entities.
  3. Select a Dialog Resumption option:
  4. (Optional) Enable Amend Hidden Entities to allow amendment of entities marked as Hidden.

Task Level

  1. Go to Conversation Skills > Dialog Tasks → open the task.
  2. Click ⋯ (More Options) > Manage Amend Behavior.
  3. Select Customize for this task: (Optional) Enable Amend Hidden Entities at the task level.

Amend Trigger Patterns

The platform recognizes these utterance patterns as amend requests. ~amend_synonyms includes: amend, change, modify, alter, update, replace, make, move, upgrade, want.
Multiple entity values can be amended in a single utterance. Example: “Change the departure date from today to tomorrow and departure city to Chicago.”