Turn long sales call transcripts into clear insights for your sales, marketing, and product teams. The flow reads one transcript, extracts structured takeaways for each team, and pushes results to your CRM and Notion so follow up is fast and consistent.

An execute workflow trigger starts the run. A prompt builder standardizes names and context, then three AI agents read the transcript for sales, marketing, and product. Each agent uses Azure OpenAI and a structured output parser to return clean fields. Set nodes capture the AI output and metadata. The flow routes sales results to a Salesforce subworkflow and sends marketing and product results to Notion subworkflows. All results are merged, bundled, and marked as complete. Built in logic supports reruns for only the items that failed, which helps with rate limits and large batches.

You will need Azure OpenAI, a Salesforce account, and Notion workspaces. Set credentials in n8n Cloud and map your fields to match your CRM and Notion databases. Teams can expect to cut review time from about 60 minutes to about 5 minutes per call while improving data consistency. Common uses include call summaries, objection tracking, campaign ideas, and product feedback capture right after each discussion.

• Execute Workflow Trigger starts the process when called by another flow
• Create User Prompt standardizes the transcript and context for all agents
• Three Azure OpenAI Chat Model nodes power separate agents for sales, marketing, and product
• Structured Output Parser nodes enforce consistent fields and formats
• Set nodes capture AI output and metadata for downstream steps
• Execute Workflow nodes route results to Salesforce and Notion subworkflows
• Merge and Bundle steps combine outputs from all teams into one package
• Rerun logic allows processing only remaining items after a failure

• Reduce manual review from about 60 minutes to about 5 minutes per call
• Send tailored insights to sales, marketing, and product in one pass
• Improve data quality with predefined structured fields for each team
• Connect Azure OpenAI, Salesforce, and Notion without manual exports
• Handle higher call volume with rerun on failure and rate limit handling

1. Import the template into n8n: Create a new workflow in n8n, click the three dots menu, select 'Import from File', then choose the downloaded JSON file
2. You will need accounts with Azure OpenAI, Salesforce and Notion. See the Tools Required section above for links to create accounts with these services.
3. In the n8n credentials manager, create an Azure OpenAI credential. Enter your Azure OpenAI endpoint and API key, and set the deployment name for gpt 4o mini. Assign this credential to the three Azure OpenAI Chat Model nodes.
4. Open the Salesforce subworkflow linked by the SF Sales Data Processor and create a Salesforce credential in n8n Cloud. Double click the Salesforce node in that subworkflow, choose Create new credential, then follow the on screen steps to connect your org.
5. Open the Marketing and Product subworkflows. Double click the Notion nodes inside each subworkflow and create a Notion credential. Approve the workspace and grant access to the target databases.
6. In the Create User Prompt node, map your incoming transcript field. If your parent workflow sends a field named transcript, ensure the expression reads it correctly.
7. Review the Structured Output Parser nodes. Adjust the schema fields to match your Salesforce objects and Notion database properties.
8. Verify Execute Workflow nodes point to the correct subworkflow IDs and that you have permission to run them.
9. Run a test with a short sample transcript. Confirm the sales, marketing, and product outputs appear in the Merge and Bundle steps.
10. Check Salesforce for an updated record and Notion for new pages. If rate limits occur, rerun the job to process only remaining items.
11. Connect your parent workflow to this one using the Execute Workflow node so calls trigger this processor. Toggle the workflow to Active when ready.