View Categories

AI Workflows

1 minutes

Introduction #

AI workflows enable the automation of intelligent processing on Kafinea records. By adding an AI Prompt task to a workflow, you can have data automatically analyzed, summarized, translated, or classified by an artificial intelligence model.

1. Create a workflow with an AI task #

  1. Access Settings > Automation and Communication > Workflows
  2. Create a new workflow or edit an existing one
  3. Choose the target module (Contacts, Invoices, Quotes, etc.)
  4. Define the trigger conditions
  5. Add an AI Prompt task

2. Configure the AI Prompt task #

The AI Prompt task consists of two parts:

The prompt (instruction) #

This is the textual instruction sent to the AI model. You can include module variables to customize the request.

Prompt example:

Résume en 2 phrases le contenu de la description suivante : $description
Classifie le niveau d'urgence du ticket parmi : faible, moyen, élevé, critique.

The return field mapping #

For each field the AI needs to fill, you configure:

  • The target field: the record field to update
  • The specific instruction: what the AI should generate for this field

Important: The AI’s response is expected in JSON format with field names as keys. The system automatically handles extraction even if the response contains text around the JSON.

3. Contextual data transmitted to the AI #

The AI task automatically transmits the following to the model:

Data Description
Record data All fields of the current record
Parent records Data from linked records (contact, account, etc.)
Module descriptions Field names and types, picklist values
Detail lines For inventory-type modules (invoices, quotes, orders)
Organization data Company information linked to the record
Document content If the module is Documents, the content of the attached file

4. Usage examples #

Automatic ticket summary #

  • Module: HelpDesk
  • Trigger: On creation
  • Prompt: “Summarize this support ticket in one concise sentence.”
  • Target field: cf_summary ← instruction: “A sentence summarizing the issue”

Contact classification #

  • Module: Contacts
  • Trigger: On save
  • Prompt: “Analyze this contact’s information and suggest a business segment.”
  • Target field: cf_segment ← instruction: “The segment from: SMB, Mid-Market, Large Enterprise, Individual”

Product description translation #

  • Module: Products
  • Trigger: On save
  • Prompt: “Translate the product description into English.”
  • Target field: cf_description_en ← instruction: “The English translation of the description”

5. Error management #

If the AI does not return a valid response, the workflow continues without updating the fields. Errors are logged in the server logs:

  • Missing API key: check the AI gateway configuration
  • Empty response: the prompt may be too vague or the data insufficient
  • JSON parsing error: the system automatically attempts several extraction strategies

Tip: Consult the PHP logs (error_log) to diagnose AI workflow execution issues.

6. Limitations #

  • The prompt size is limited by the AI model used (generally 4,096 response tokens max).
  • Large image files or very long documents may be truncated.
  • The workflow runs synchronously: a slow AI call can slow down processing.

Related References #