Quick overview
This workflow runs weekly (or manually) to collect recent customer feedback from Salesforce Cases and Jira tickets, redact and deduplicate it, classify it with an OpenAI-compatible LLM, then cluster and route issues by category by creating or updating Jira VOC issues and creating Salesforce follow-up tasks.
How it works
- Runs on a weekly schedule (or on demand) and computes a lookback time window for collecting new feedback.
- Pulls recent Salesforce Cases, Jira issues for a configured project, and Salesforce Account data (or uses built-in sample feedback when enabled).
- Normalizes the text, redacts emails and phone numbers, drops duplicates using workflow static data, and flags VIP items based on account revenue, rating, or an allowlist.
- Sends the batch of feedback to an OpenAI-compatible chat model (configured for Featherless AI) to classify each item by category, theme, sentiment, urgency, and churn risk.
- Aggregates the classified feedback into category clusters, assigns priority, and builds an HTML weekly digest summary.
- When live mode is enabled, routes clusters by team by commenting on or creating labeled VOC issues in Jira for product topics and creating Salesforce follow-up Tasks for billing, onboarding, sales/engagement, and escalations.
Setup
- Add credentials for Salesforce and Jira Software Cloud, and an OpenAI-compatible API credential for the LLM node (update the base URL/model if you are not using Featherless AI).
- In the Config values, set your Jira project key, Jira project ID, and Jira issue type ID, and decide your lookback period, VIP rules, and max items per run.
- Set use_sample_data to false to pull from Salesforce and Jira, then set dry_run to false to allow the workflow to create Jira issues/comments and Salesforce Tasks.
- (Optional) Enable and configure Google Sheets logging by setting the spreadsheet ID and ensuring a “Feedback Log” sheet exists.
- (Optional) Enable the Gmail node and provide Gmail credentials, then set the recipient email to send the weekly digest.