Hey everyone! I’m trying to build something cool but I’m stuck and need some help.
My goal: I want to say “Hey Google, talk to AI assistant” and have my Google Home send that question to my self-hosted n8n server. Then n8n would call the OpenAI API, get a response, and send it back so Google can speak the answer out loud.
Current status: Not working at all. Google Home isn’t connecting to my n8n setup.
What I’ve done:
Created an n8n workflow
Got my OpenAI API key ready
Set up Google Cloud console account
Installed SSL certificate using certbot
Opened ports 80 and 443 for n8n access
Partially completed Google’s OAuth2 application form (left some fields empty but it accepted submission)
Generated OAuth2 credentials in n8n
Tried adding them to Google but it’s asking for scopes and I’m lost
Questions:
Which OAuth scope should I use?
What are the exact steps to make n8n respond to Google Home voice commands?
Is this still doable? I see old tutorials mentioning Dialogflow but that service ended in June 2023
I thought this would be a common project but I can’t find recent guides. Any working tutorials or step-by-step instructions would be awesome!
Google Actions Console is a mess right now and their docs are terrible. Just use Tasker on Android - it’s way simpler. Set up a voice command in Tasker that calls your n8n webhook directly, then use Tasker’s TTS to read back the OpenAI response. No OAuth hassles at all.
Went through this exact setup 6 months ago and hit the same walls. OAuth is way too complex for what you need. Here’s what actually worked: Google Assistant routines + IFTTT as a bridge. Set up an IFTTT applet that triggers on your phrase, then sends a webhook to your n8n instance. Your n8n workflow catches the webhook, runs it through OpenAI, and returns the response. For audio playback, I used Google’s Text-to-Speech API to generate audio files served back through the webhook response. Takes about 3-4 seconds end to end. Way more reliable than wrestling with Google’s developer console authentication mess.
Google deprecated direct webhook access for custom actions, so your n8n setup is hitting their restrictions. All that OAuth pain exists because they want you using enterprise solutions.
I ran into this same issue building voice automation for our office. Wasted days on SSL certificates and webhook timeouts.
Switching to Latenode solved everything. They’ve got native Google Assistant integration that skips the OAuth nightmare entirely. No SSL setup, no port forwarding, no webhook debugging.
The workflow’s simple: voice trigger hits their Google Assistant node, goes straight to their OpenAI connector, returns speech. Takes 20 minutes to build.
Best part? Sub 2-second response times since it runs on their cloud. No more self-hosting headaches or wondering if your connection dropped the webhook.
I use it for everything now - way more reliable than fighting Google’s constantly changing APIs.
Been there with n8n workflows getting messy fast. OAuth is a nightmare - you’ll waste weeks debugging auth issues.
You’re overcomplicating this. Skip Google’s developer console and OAuth headaches. Build everything in one place that actually works.
I switched to Latenode for this exact project last year. Native Google Assistant integration without OAuth pain, plus OpenAI connectors built in. Just drag and drop.
Super simple workflow: voice trigger → OpenAI call → text response. No webhook debugging, SSL certificates, or port forwarding hell.
30 minutes and I was done vs weeks fighting n8n and Google’s APIs. Faster response times too since it runs on their infrastructure.
Got this working with n8n two months ago after some trial and error. Google Home won’t talk directly to custom webhooks anymore - you need their Actions SDK or a third-party bridge. Don’t bother with OAuth, that’s for different stuff. I used Home Assistant as middleware. Set up a custom intent in Home Assistant to catch your voice command, then it fires an HTTP request to your n8n webhook. n8n processes it through OpenAI and sends the response back to Home Assistant, which speaks the answer through TTS. You’ll need Home Assistant running with n8n, but it handles all the Google Home integration pain. Way cleaner than wrestling with Google’s developer console and more reliable than IFTTT’s delays.
Had the same headaches until I found something that actually works. Google killed direct webhook support for custom actions - that’s why n8n won’t play nice. Those OAuth scopes (assistant-sdk-prototype or actions.devices.SYNC) are deprecated and unreliable anyway. Try Node-RED with the node-red-contrib-google-assistant package instead. It skips Google’s auth nightmares and hooks straight into Assistant SDK. You’ll still need SSL, but the setup’s way cleaner. Just build a flow that grabs voice input, sends it to OpenAI through an HTTP request node, then pipes the response back through the assistant node. Gets consistent 2-second responses without fighting Google’s ever-changing API mess.
dialogflow’s still here, just a new name - conversational ai. for simple stuff, forget oauth, just use webhook triggers in n8n, then point google actions to your public endpoint. way simpler than messing with oauth!