Automating Zoom Meeting Insights into HubSpot with n8n

Sales conversations contain valuable information, but when Zoom transcripts sit in Google Drive as unstructured PDFs, turning those insights into useful CRM data can become a manual, time-consuming task.

I designed an n8n workflow to automate that process, from transcript extraction to HubSpot contact and deal updates.

How the workflow works

Google Drive retrieves Zoom transcript PDFs.

PDF Extract converts documents into raw text.

Regex filtering removes unnecessary content that can interfere with extraction.

Qwen AI identifies company names, job titles, pain points, objections, and next steps, returning structured JSON.

n8n cleans and maps the extracted fields.

HubSpot creates or updates the relevant contact and deal records.

The main challenge

AI can sometimes misinterpret compound job titles, such as “VP of Sales & Marketing,” as two separate people. I’m refining the extraction prompt to improve accuracy and handle ambiguous roles more reliably.

Why use n8n?

Instead of relying entirely on rigid, off-the-shelf integrations, this approach provides greater control over AI prompts, data mapping, error handling, and CRM workflows. It helps reduce repetitive data entry while keeping important sales insights organised.

What’s next?

I’m exploring Gemini as a fallback model when key fields are missing and investigating sentiment analysis to extract additional context from sales conversations.

This workflow could be useful for sales teams, founders, and growing businesses that want to streamline meeting follow-ups and maintain cleaner HubSpot records.

Tools: n8n, Google Drive, Zoom transcripts, Qwen AI, HubSpot, PDF extraction, and regex filtering.

Are you using AI to turn meeting transcripts into structured CRM data? I’d be interested to hear what’s worked for you.

#n8n #WorkflowAutomation #AIAutomation #HubSpot #SalesOperations