Vapi doesn’t have built-in persistent memory, but you can set it up by managing conversation history externally and feeding that context back into your assistant when needed. After each call, send the call data to your make(can be n8n or zapier) webhook to send the call data such as transcripts, summaries, or key points to an external storage. You can use something simple like Google Sheets or Airtable through [Make.com](
http://Make.com), or something more advanced like database or vector database if needed. They both will work. Once the data is stored and linked to each user, usually by phone number or user ID, you’ll want to retrieve that information when the user calls back or when you initiate an outbound call. So basically for outbound example, it will see that number and look at the information about that number. Make sure to add the information dynamically through the prompt. By dynamic I mean 'the person calling's name is {{callername}}.' In your prompt and the name will change dynamically based on who is calling. The basic workflow is: Vapi sends call data to [Make.com](
http://Make.com), Make stores it, and when a new call happens, Make retrieves the history and injects it back into Vapi.