So many people are just weird about speaking to a voice agent. We have tried several different greetings to build confidence, but people still simply want to leave a voicemail. For those of you building successful voice agents to answer the phones. What have you found to be the best greeting to lead people to a conversation?

We are currently using a platform called CoStar to look up commercial property info. The monthly fee is $600 for one person! So I found a platform that has all the commercial data we need, and built a lookup tool in Loveable. It has a Supabase backend to store the properties, has account handling for secure login, includes buttons to launch Google Maps or Goole Earth. I used a combination of Claude to build my workflow in n8n and lovable. There was some back and forth getting everything working, and getting the data set up so Lovable could use it, but it only took me a few hours after creating the workflow. Version 2 will include the ability to build a quote by selecting the services we offer and a price, then passing it all to a template in PandaDoc. From there it can be sent to a potential client where they can select which services they want, and e-signing the proposal. Really excited for this!

Sharing is caring. This is a very simple workflow that reports errors to you, AND tells you what to fix. It has saved me many hours trying to figure out what was wrong. It uses n8n itself to debug your workflows. It costs nothing, and will point you to the exact problem with your workflows. Simply replace your n8n key and your ‘send from’ email, and you’ll get immediate notification when a workflow fails, a detailed description, and suggestions of exactly how to fix it. After setup, just go to the “Settings” in each workflow and under errors, select this workflow. Whenever/if it fails, this workflow will be triggered. Here is an example of the email you receive, with a link directly to the execution that failed. Hello, there has been an error in a workflow. Please check below for details: Direct Link To Affected Execution: https://xxxxxxxxxxxxxxxxxxxxxxxxx Name Of Affected Node: Clean up email Type Of Affected Node: n8n-nodes-base.set Error Message: Node 'Get Call Data' hasn't been executed 🤖 AI Debugging Assistant Analysis: Root Cause Analysis: The error "Node 'Get Call Data' hasn't been executed" occurs because of a logical branching issue in the workflow. The Check for CallID (IF node) creates two distinct execution paths. True Path: If a callId exists, the workflow proceeds to the Get Call Data (HTTP Request) node to fetch additional information. False Path: If a callId does not exist, the workflow bypasses Get Call Data and connects directly to the failed node, Clean up email. The Clean up email (Set node) contains an expression for the phone field that explicitly tries to access data from the Get Call Data node: $('Get Call Data').item.json.customer.number. When the workflow takes the "False Path" (because there is no callId), the Get Call Data node is skipped. Consequently, when the Clean up email node tries to execute its phone expression, it fails because Get Call Data has no output data, leading to the reported error.

I built a full AI voice receptionist using VAPI on the front-end and n8n on the back-end, powered by an n8n MCP server that connects to seven n8n workflows. In this video, I walk through the entire system from scratch and show how the agent gathers info, manages the CRM, books and updates appointments, transfers calls, answers questions, logs every call, and keeps an appointment history. Everything is built with no code and designed to be easy to customize for your own use case. By the end, you’ll understand the full setup and have everything you need to launch your own AI receptionist. GOOGLE SHEET TEMPLATE

I have two transient agent workflows that have disappeared from my cloud-hosted n8n account. They are not showing up as archived. The weird thing is that they are still working when I call the number. Any guidance would be greatly appreciated!