Build an AI-Powered CRM Assistant with HubSpot, n8n & ChatGPT

Manually chasing leads is exhausting. You send an email, they reply with a question, and now you’re stuck in a back-and-forth, checking your inbox every few hours so you don’t miss anything.

What if you could have a smart assistant that instantly replies to leads, personalises responses, and logs every interaction in your CRM… without you touching a thing?

In this tutorial, we’ll build an AI-powered CRM assistant that connects HubSpot, n8n, and ChatGPT. You’ll end up with a system that:

• Captures form submissions from HubSpot
  
• Sends the message to ChatGPT for a smart reply
  
• Logs the AI’s response directly on the lead’s profile in HubSpot
  
• Is easy to expand into SMS, email replies, or other channels later
  

And the best part? We’ll do it all in one weekend.

What You’ll Need

• HubSpot CRM (free account or developer test account)
• n8n (free cloud account or self-hosted)
• OpenAI API key (gives access to GPT models)
• Basic familiarity with forms, workflows, and APIs (helpful but not required)
  

Step 1: Create a HubSpot Form

Our assistant needs something to trigger on, so we’ll start with a form where leads can send messages.

1. Go to Marketing → Forms in HubSpot.
   
2. Click Create Form → Choose Embedded Form (or Standalone for testing).
   
3. Add these fields:
   
• First Name → maps to firstname
• Last Name → maps to lastname

• Email Address → maps to email

• What do you need help with? → create a new contact property called service_inquiry.
  
4. Publish the form and save its share link.
   

💡 Important: Ensure every form field is mapped to a contact property- ensures we can send the data to n8n later.

Image: Huspot form with a new property created

Step 2: Send Form Data to n8n via a HubSpot Workflow

HubSpot will handle sending the form details to our automation.

1. Go to Automation → Workflows.
   
2. Click Create Workflow → Start from scratch → Contact-based.
   
3. Set the trigger:
   
• Form submission → Select the form you just created.
  
4. Add an action:
   
• Trigger a Webhook → Method: POST

• URL: Your n8n webhook URL (we’ll set this up in Step 3)
  

Request body:

{
  "contact_id": "{{ contact.hs_object_id }}",
  "first_name": "{{ contact.firstname }}",
  "email": "{{ contact.email }}",
  "message": "{{ contact.service_inquiry }}"
}

Save and activate the workflow.

Image: The Hubspot Outgoing webhook

Step 3: Build the n8n Flow

This is where the magic happens. We’ll take the webhook data, run it through ChatGPT, and send the reply back to HubSpot.

1. Webhook Node

• Add a Webhook node in n8n:
  
• Method: POST

• Path: incoming-lead-message


Respond immediately with:

{ "status": "received" }

Image: n8n node that handles incoming webhook

2. ChatGPT Node

• Add an HTTP Request node connected to the webhook.
  
• Method: POST

• URL: https://api.openai.com/v1/chat/completions

• Headers:
  
• Authorization: Bearer YOUR_OPENAI_API_KEY

• Content-Type: application/json


Body:

{
  "model": "gpt-3.5-turbo",
  "messages": [
    {
      "role": "system",
      "content": "You are a helpful assistant responding to leads for a marketing agency."
    },
    {
      "role": "user",
      "content": "{{$node['Webhook'].json.body.message }}"
    }
  ]
}

Note: Feel free to refine the prompt to better fit your use case. Also, you can switch to the GPT model that best meets your needs.

Image: ChatGPT Reply node

3. HubSpot API Node (Log Reply)

• Add another HTTP Request node.
  
• Method: POST

• URL: https://api.hubapi.com/crm/v3/objects/notes

• Headers:
  
• Authorization: Bearer YOUR_HUBSPOT_ACCESS_TOKEN

• Content-Type: application/json


Body:

{
  "properties": {
    "hs_note_body": "{{ $node['ChatGPT Reply'].json['choices'][0]['message']['content'] }}",
    "hs_timestamp": "{{new Date().toISOString()}}"
  },
  "associations": [
    {
      "to": {
        "id": "{{$node['Webhook'].json.body.contact_id }}"
      },
      "types": [
        {
          "associationCategory": "HUBSPOT_DEFINED",
          "associationTypeId": 202
        }
      ]
    }
  ]
}

Note: This example assumes you are creating a contact note in Hubspot. If you want to handle any other entity, please refer to the Hubspot API for the right endpoint and the association type IDs.

Step 4: Test the Workflow

1. Open your HubSpot form and submit a test entry.
   
2. In n8n, watch the Webhook node capture the payload.
   
3. ChatGPT should generate a reply.
   
4. The HubSpot API node logs that reply as a Note on the contact’s profile.
   

Check your contact in HubSpot, you should see something like:

Step 5: Ideas for Improvement

This is just the foundation. You can:

• Send replies via email or SMS instead of just logging notes
  
• Add keyword triggers to escalate to a human
  
• Personalise tone based on lead source or past conversations
  
• Turn it into a Hubspot Marketplace app if you want to commercialise it
  

Wrapping Up

In a weekend, you’ve gone from a blank CRM to a smart assistant that replies instantly to leads, never forgets to follow up, and logs everything neatly.

There you go! This should be a good starting point for even more ambitious AI-driven sales tools.

If you’d like this exact workflow set up for your business, or customised to your sales process, book a discovery call with me.