Unfortunately, I couldn't attend the build hours call this morning but wanted to thank Alejo for our one-on-one classroom session yesterday. If your goal is to troubleshoot a current project or to get up to speed quicker, I highly recommend booking a session. Alejo was able to assist me with 3 different workflows in that time!

I have a general discussion question for the group. Looking for your thoughts and ideas on this one. I've heard a few people mention that their AI is learning and getting better at responding based on the history of calls it has made. Meaning it is able to reference past calls to effect the way it answers future calls. Improving the quality of responses based on the feedback from actual calls. From my understanding using services like VAPI every call is a world to it's own. We can feed it previous conversation but this would mostly be applied to the same caller calling back... as opposed to the complete history of calls. Are we able to get our AI to grow and get better? Learn from great previous responses or great flow state calls? Or is this heading into the realm of custom LLM? As I said, I have heard people mention this out there in the AI realm... are they give us BS or is this something that we can also work towards?

Hello All! My son and I are new to this and working on an outbound agent with Vapi and Make. The Make scenario is running successfully but we don't receive a call. In Vapi, under advanced settings in the agent, there's a Server URL and secret. Do we need our own server, or do we use a URL created with an HTTP module in Make? Any troubleshooting advice or fields I should look at in Vapi are greatly appreciated! **Updated** Turns out in the HTTP module it needed the agent ID vs agent.