Capture client conversations, turn them into short notes, and alert the right person fast. A clean summary also lands in your CRM so your team always sees the latest context. This fits support and customer success teams that handle many calls or meeting transcripts.

Data enters through a web form that asks for the client email and the conversation. An AI model writes a two to three sentence summary. The flow searches HubSpot by the email and adds a meeting note on the contact with that summary. A routing agent reads the full text, picks one owner from your preset list, and uses Gmail to send a focused update. A final screen shows what was sent and to whom.

Setup needs accounts for OpenAI, HubSpot, and Gmail. Add your team addresses in the routing list and make sure the names in the agent prompt match those labels. Expect faster handoffs, fewer mistakes, and a full record after every client touchpoint. Common uses include support escalations from call transcripts, onboarding feedback, and product issue reports. Many teams cut triage from 30 minutes to 3 minutes per case with this approach.

• Form capture for client email and conversation text using a hosted form
• AI summarization that condenses transcripts into two to three sentences
• HubSpot contact search by email and automatic meeting note creation
• Configurable routing list where you set the exact team email addresses
• Routing agent selects one owner and composes a focused Gmail message
• Gmail tool sends the email from your connected account
• Completion screen shows the agent output and recipient for quick review
• Secure credentials with OAuth2 for HubSpot and Gmail and API key for OpenAI

• Reduce manual triage from 30 minutes to 3 minutes per case
• Streamline owner handoff by up to 80 percent
• Improve CRM data completeness with summary notes on every contact
• Connect HubSpot and Gmail in one smooth flow
• Handle up to 5 times more client notes per day with the same team
• Cut misrouted emails by using a clear single owner decision

1. Import the template into n8n: Create a new workflow in n8n > Click the three dots menu > Select 'Import from File' > Choose the downloaded JSON file.
2. You'll need accounts with HubSpot, Gmail and OpenAI. See the Tools Required section above for links to create accounts with these services.
3. Open the OpenAI Chat Model node. In the credential dropdown, click Create new credential. Add your OpenAI API key from the OpenAI API page and save. Select this credential for the node.
4. Open the HubSpot nodes. For each one, in the credential dropdown, click Create new credential, choose HubSpot OAuth2, and follow the sign in steps. Save and select the same HubSpot credential for both HubSpot nodes.
5. Open the Gmail tool node. In the credential dropdown, click Create new credential, choose Gmail OAuth2, approve the requested scopes, then save and select this credential.
6. Open the node named Define routing emails here. Replace the sample addresses with the real owners for support, product, success, or admin. Save your changes.
7. Open the Router Agent node. Review the system message options and make sure the labels match the names you used in the routing list. Adjust wording so the agent can pick exactly one owner.
8. Open the HubSpot search node and confirm the filter uses the email field. Keep the association in the HubSpot meeting node so the note attaches to the found contact.
9. Activate the workflow. In the Enter Client Transcript trigger node view, open the form URL in a new tab. Submit a test with your own email and a sample conversation.
10. Check HubSpot for the contact. You should see a new meeting note with the AI summary attached to that contact.
11. Check the inbox of the selected owner for the Gmail alert. If no email arrives, confirm Gmail scopes, sender account, and spam folder.
12. Review the completion page output. It should show who was notified and the short message.
13. Troubleshoot common issues: if no HubSpot contact is found, verify the email exists in HubSpot. If routing seems off, refine the agent prompt or the option names. If OpenAI fails, confirm the API key and selected model.