- For build 3, at the end, I created Web-hook solution for both Make.com & n8n to see what feels more natural to me. - For build 4, I skipped first part with Liam voice app and do similar solution with Voiceflow & Meta WhatsApp Cloud API. Week 2 on the way.

Finally got my first real AI voice receptionist workflow live. Built with Vapi + n8n to handle inbound calls, check availability, and book appointments straight into Google Calendar. Would love feedback from anyone who’s built something similar or ideass for what I should add next in V2.

Build 3 is on the way!

Just finished my AI automations course! Below are screenshots of all three successful workflows.👇 Not everything went smoothly — I ran into errors and had to debug along the way. But fixing those issues was honestly the best part! That’s when things really started to click and connect.😌 Feels good to see everything working in the end. Excited to keep building and learning. Lets get it!!! 💪

It is working as planned. 😀 If i got i right, I saw that every time you run a tool or an agent on "Relevance AI" it uses "Actions" and "Credits" (See attached picture). I think I have found a workaround by selecting for LLM(brain) "Mistral: Devstral 2 2512 (free)". It is a open‑source large language models that Relevance AI runs on their own infrastructure and it doesn't use "Actions" and "Credits". For me this is acceptable in this phase of learning. For first "paid" job i will use Gemini, Open AI, etc. I hope this post will help someone so they can experiment and stay in free (non-paid) relevance ai subsriction.