The goal is for the bot to remember the returning clients. Lets take the scenario of a pizza place. Mary orders a cheese pizza with a can of coke. The bot gets this info and sends it to the POS of the pizza joint but stores this in the database. Lets say Mary calls the same place, one week later. Upon recieving the call from Mary (perhaps the bot matches the phone number), it loads in the old context of the call it had with Mary and instead of taking the order as programmed. It dynamically changes the prompt. For e.g It might ask Mary if she would like the same order as before and if she confirms, it adds the cheese pizza with a can of coke to the POS of the restaurant. This is something I am trying to make work. Would be great if this could be done inside the Vapi platform.

@DodoBird Absolutely, the scenario you've described is a great example of personalized customer service, and it's something that can be implemented within the VAPI platform. By utilizing a combination of voice recognition, caller ID, and a customer relationship management (CRM) database, the system can recognize returning clients like Mary and recall their previous orders. When Mary calls again, the system would identify her phone number, retrieve her order history from the database, and the bot could then ask if she would like to repeat her last order. If Mary confirms, the bot would proceed to place the order for a cheese pizza with a can of coke to the POS system of the pizzeria. To implement this, you would need to ensure that the VAPI platform is integrated with a CRM system that can store and retrieve customer order histories. Additionally, the voice bot would need to be programmed to handle the dynamic prompt change based on the retrieved context. If you need more detailed guidance on setting up this system or integrating it with your existing infrastructure, please ask support for more info on this.

tell me more about the crm system.

@DodoBird To integrate a CRM system with VAPI for the scenario you've described, you would typically follow these steps: 1. **CRM Selection**: Choose a CRM system that supports API access for creating, updating, and retrieving customer data. 2. **Database Structure**: Set up the CRM database to store customer profiles, including contact information and order history. 3. **VAPI Integration**: Develop or configure the VAPI platform to interact with the CRM's API. This would involve: - Authenticating with the CRM system. - Fetching customer data based on the caller's phone number. - Storing new order information after each call. 4. **Voice Bot Customization**: Customize the voice bot's script to: - Recognize returning customers using caller ID. - Retrieve the customer's last order from the CRM. - Ask the customer if they would like to repeat their last order or place a new one. 5. **Order Processing**: If the customer confirms they want the same order, the bot would send the order details directly to the POS system. If the customer wants a new order, the bot would proceed with the standard ordering process. 6. **Data Synchronization**: After the call, synchronize the new order details back to the CRM system to update the customer's order history. 7. **Testing**: Thoroughly test the system to ensure that the integration works seamlessly, and the bot correctly interprets and responds to customer requests. 8. **Maintenance and Updates**: Regularly maintain the system to handle any API changes from the CRM provider and to improve the customer interaction based on feedback. Please note that this is a high-level overview, and the actual implementation may require specific technical steps depending on the CRM system and VAPI platform you are using. If you need technical assistance or more detailed instructions, please ask support for more info on this.

@DodoBird You need to append the context into assistant messages with role of type system and instruct using your prompt, on how to use the previous call contexts.

json
{
  "model": {
    "provider": "openai",
    "model": "gpt-4o",
    "messages": [
      {
        "role": "system",
        "content": "insert-system-prompt-here"
      },
      {
        "role": "system",
        "content": "insert-previous-call-context-here"
      }
    ]
  }
}

You the man @Shubham Bajaj. Thanks. Wlyk.

@DodoBird do update me how it goes for you.