How to instructions for adding object attributes for AI Agents: ticket-categorizer and ticket-text-extractor

Technical assistance
1

Infos:

  • Used Zammad version: 7.1
  • Used Zammad installation type: package
  • Operating system: debian
  • Browser + version: Firefox 152

Expected behavior:

creating AI agent ticket extractor or AI agent ticket categorizer . see docs AI Agents — Zammad Admin Documentation documentation documentation should display step by step instructions and/or example screenshots for required settings to successfully enable ticket extractor or ticket categorizer . Zammad user expects simple steps, instructions, and/or screenshots to enable and properly configure AI agent with object attributes Objects — Zammad Admin Documentation documentation
*

Actual behavior:

configuring AI agent ticket extractor second setup config page shows red error box “No suitable object attributes found, please add one of the supported types first: input, textarea, integer, select, tree_select”

configuring AI agent categorizer second setup config page shows red error box : "No suitable object attributes found, please add one of the supported types first: select, multiselect, tree_select, multi_tree_select "

Zammad user confused how to configure page /#system/object_manager to match requirements on page /#ai/ai_agents

Documentation does not cover how-to setup object attributes for two AI agents: ticket text extractor and ticket categorizer

Documentation pages

Steps to reproduce the behavior:

Zammad with no new correctly created attribute objects
*

2

Not sure if I got you right. The text of both AI agents says clearly that an object attribute is required and even links to the object attribute section:

Categorizer:

Because there is no built in category field in Zammad, you first have to create a custom ticket attribute in which you want to track the ticket’s category. The field type must be one of the select types.

Text extractor:

Make sure to have a custom ticket attribute in which you want to store the extracted information. Otherwise, create a suitable > one before configuring the AI agent. The field type must be one of the following:

  • Text
  • Textarea
  • Integer
  • Single selection
  • Single tree selection

For select-type attributes, the wizard allows you to limit the options the AI agent can choose > from and to add descriptions for them. For these types, the AI agent also receives the list of > allowed options to choose from in case of a match.

And to avoid duplicate content, there are just links to the object attribute section.

At least for the categorizer agent, a scenario is that you already have a custom category field in which you track the categories manually.

2 Likes