Quick Guide: Create Custom GPTs with AI (Beginner Tutorial)

Ever wanted to create your own AI assistant tailored to your specific needs? With OpenAI’s new GPTs, you can do just that – and it’s easier than you might think. In this guide, I’ll walk you through creating a custom GPT from scratch, complete with its own knowledge base and custom actions. Let’s dive in!

What We’re Building

We’re going to create a solar sales assistant GPT. This AI helper will:

1. Assist junior sales reps with company procedures
2. Provide roof size estimates
3. Calculate potential solar savings for customers

Step 1: Setting Up Your GPT

1. Go to the ChatGPT homepage
2. Click the “Create” button
3. Select “Configure” to start from scratch

Step 2: Basic Configuration

1. Name your GPT (e.g., “Smith’s Solar Sales Assistant”)
2. Add a description (e.g., “A sales rep assistant for Smith’s Solar Company”)
3. Write instructions for your GPT’s behavior

Here’s a sample instruction set:

You are Smith's Solar Sales Assistant, designed to help junior sales reps learn company procedures and provide solar estimates. You can:

1. Answer questions about company policies and sales processes
2. Estimate roof sizes for potential customers
3. Calculate potential solar savings using our custom solar calculator

Always ensure CRM updates are accurate and prompt.

Step 3: Adding Knowledge

1. Scroll down to the “Knowledge” section
2. Upload your company’s sales training document
3. Make sure all options are selected for document parsing

Step 4: Creating a Custom Action

This is where things get exciting. We’re going to add a custom solar calculator function:

1. Go to the “Add actions” section
2. Click “Add action”
3. In the schema area, paste the following code:

{
  "openapi": "3.0.0",
  "info": {
    "title": "Solar Calculator API",
    "description": "Calculates solar potential based on roof size and current power usage",
    "version": "1.0.0"
  },
  "servers": [
    {
      "url": "https://your-replit-url-here.repl.co"
    }
  ],
  "paths": {
    "/calculate": {
      "get": {
        "summary": "Calculate solar potential",
        "parameters": [
          {
            "name": "roof_size",
            "in": "query",
            "required": true,
            "schema": {
              "type": "number"
            }
          },
          {
            "name": "monthly_bill",
            "in": "query",
            "required": true,
            "schema": {
              "type": "number"
            }
          }
        ],
        "responses": {
          "200": {
            "description": "Successful response",
            "content": {
              "application/json": {
                "schema": {
                  "type": "object",
                  "properties": {
                    "estimated_savings": {
                      "type": "number"
                    },
                    "recommended_system_size": {
                      "type": "number"
                    },
                    "estimated_cost": {
                      "type": "number"
                    },
                    "payback_period": {
                      "type": "number"
                    }
                  }
                }
              }
            }
          }
        }
      }
    }
  }
}

1. Replace “your-replit-url-here” with the URL of your Replit project (we’ll set this up next)

Step 5: Setting Up the Backend

1. Go to Replit.com and create a new project
2. Fork the solar calculator template (available in the resource hub mentioned earlier)
3. Click “Run” to start your backend server
4. Copy the URL of your Replit project

Step 6: Finishing Touches

1. Add an image to your GPT using DALL-E
2. Save your GPT (you can choose to make it public or keep it private)

Testing Your Custom GPT

Now that everything’s set up, let’s test it out:

1. Ask about company offers:
   “What offers do we have to close a sale?”
2. Use the custom solar calculator:
   “A roof is 120 square meters and the family spends $450 per month on power.”

Your GPT should now be able to answer questions about company policies and provide solar estimates using the custom calculator!

Wrapping Up

Congratulations! You’ve just created a custom GPT with its own knowledge base and a powerful custom action. This is just scratching the surface of what’s possible with custom GPTs. As you get more comfortable, you can add more complex actions, integrate with other APIs, and create truly powerful AI assistants tailored to your specific needs.

Remember, the key to a great custom GPT is high-quality data and well-defined actions. Keep refining your knowledge base and custom functions, and you’ll have an incredibly useful AI helper in no time.

Happy GPT building!

Need help with AI-powered automations for your business? Check out Alacranlabs.com for expert assistance.