n8n setup: build forms, support agents and webhooks

This beginner tutorial builds three n8n workflows on a Hostinger VPS. The presenter selects KVM2 with two cores and 8 GB of RAM, which he says is sufficient for the examples. The setup uses Hostinger's n8n application installer and an owner account. Although the workflow service is self-hosted, the support example calls OpenAI's API rather than running a model on the server.

The first workflow sends a guide through Gmail after a form submission and emails the owner a notification. It explains triggers, pinned test data and dynamic fields. Connecting Google requires a Google Cloud project and OAuth credentials. Publishing makes the production form available beyond the editor.

The second workflow starts as a chat-based AI agent using GPT-5 mini. The presenter explains that API billing is separate from a ChatGPT subscription. He adds simple memory and a system message containing studio facts, then replaces the chat trigger with incoming Gmail messages. Memory uses the sender's email address as its session ID. In his test, the reply includes extra details beyond the question, which he proposes addressing in the prompt.

The final workflow receives website waitlist submissions through a POST webhook and appends rows to Google Sheets. It covers matching HTTP methods, replacing the test URL with the production URL, and error notification workflows. The closing section explains HTTP request nodes for apps with an API.