Affiliate Disclosure: This guide contains affiliate links to n8n Cloud (30% commission for 12 months) and OpenRouter (20% recurring commission, 90-day cookie). Every tool in the $0 stack has a genuinely free tier — no credit card required. Affiliate links are marked with ↗.
This post is part of AIGetFree's ongoing coverage. For related content, browse our Blog or check our API Gateway Showdown.
The short version: $0 AI Automation Stack — five free tools wired together into one production-ready pipeline, tested and verified.
Wired Together, Costs Nothing
Build a Production AI Pipeline That Runs on $0/Month
n8n, free OpenRouter models, Supabase, and a Telegram bot — combined into one automated workflow that would cost $98.74/mo with paid alternatives.
MONTHLY BILL — $0 STACK#0001
n8n Community Edition$0.00
OpenRouter (26+ free models)$0.00
Supabase (500MB Postgres)$0.00
Telegram Bot API$0.00
RSS Feed Read (native)$0.00
Total / month$0.00
RSS──▶
n8n──▶
OpenRouter──▶
Supabase──▶
Telegram
no credit card · 15-minute setup · runs forever
$0Total Monthly Cost
5Free Tools Combined
∞n8n Executions
26+Free AI Models
500MBFree Database
~15minSetup Time
01 · Overview
The $0 Stack: What You're Building
Most automation guides start with "sign up for Zapier ($19.99/month)" and escalate from there. By the time you've added AI, storage, and monitoring, you're looking at $50–100/month before you've automated a single thing.
This guide builds the same stack for $0. Not a free trial, not free for 14 days — genuinely free-forever tiers that are generous enough for real production work. The trick is combining tools that each have strong free tiers into one pipeline where the whole is greater than the sum of its parts.
By the end, you'll have a fully automated workflow that monitors an RSS feed, classifies articles with AI, summarizes the best ones, stores them in a database, and notifies you via Telegram — all running on autopilot, all for $0/month.
02 · Components
The 5 Free Tools (And Why Each One)
03 · Signal Path
The Workflow: What We're Building
A fully automated content monitoring pipeline. Here's the flow:
⏰ Schedule Trigger
Fires every hour. Checks RSS feeds for new content.
📡 RSS Feed Read
Pulls latest articles from 3–5 feeds. Filters to last hour only.
🤖 AI Classifier (OpenRouter Free)
Uses Llama 3.3 70B (free) to classify each article by relevance. Returns JSON with category and confidence score.
🚦 IF Filter
Only articles with confidence > 70% and a relevant category pass through.
🤖 AI Summarizer (OpenRouter Free)
Uses Gemma 4 31B (free) to generate a 3-sentence summary of each relevant article.
📊 Supabase Insert
Stores article title, URL, summary, category, confidence score, and timestamp in PostgreSQL.
📱 Telegram Notification
Sends a formatted message with the article title, summary, and link.
04 · Build
Step-by-Step: Build It in 15 Minutes
1
Install n8n via Docker (One Command)
docker run -d --name n8n --restart always -p 5678:5678 -v n8n_data:/home/node/.n8n docker.n8n.io/n8nio/n8n
Need a VPS? Hetzner CX22 runs ~$4/month. If you have an old laptop or Raspberry Pi 4 sitting around, that works too — n8n runs on anything with Docker.
2
Create Your Free Accounts (5 Minutes)
You need three free accounts. None require a credit card:
OpenRouter
Sign up at openrouter.ai → get API key → use free models (no credits needed)
Supabase
Sign up at supabase.com → create project → get API URL + anon key
Telegram
Message @BotFather → /newbot → name it → get bot token
3
Set Up Credentials in n8n
Go to Settings → Credentials → Add Credential. Create these:
Type: OpenAI API
Base URL: https://openrouter.ai/api/v1
API Key: sk-or-v1-YOUR_KEY
Type: Postgres
Host: db.xxxxx.supabase.co
Database: postgres
User: postgres
Password: YOUR_DB_PASSWORD
Type: Telegram API
Access Token: YOUR_BOT_TOKEN
4
Create the Supabase Table (30 Seconds)
In the Supabase SQL Editor, run this:
CREATE TABLE articles (
id BIGSERIAL PRIMARY KEY,
title TEXT NOT NULL,
url TEXT NOT NULL,
summary TEXT,
category TEXT,
confidence INTEGER,
created_at TIMESTAMPTZ DEFAULT NOW()
);
5
Build the Workflow in n8n (7 Nodes)
Create a new workflow. Add these nodes in order:
Node 1
⏰ Schedule Trigger
Interval: every hour, or cron 0 * * * *
Node 2
📡 RSS Feed Read
Add 3–5 feed URLs. Enable "only items newer than 1 hour."
Node 3
🤖 AI Classifier
OpenAI Chat Model → OpenRouter → meta-llama/llama-3.3-70b-instruct:free
Node 4
🚦 IF Filter
Condition: confidence > 70 AND category != "irrelevant"
Node 5
🤖 AI Summarizer
OpenAI Chat Model → OpenRouter → google/gemma-4-31b:free
Node 6
📊 Supabase Insert
Postgres node → insert row with title, url, summary, category, confidence.
Node 7
📱 Telegram Send
Formatted message with title, summary and link.
6
The AI Prompts (Copy-Paste Ready)
Paste these into the System Message field of each AI node:
🤖 Node 3 — Classifier System Prompt
You are a content classifier. Analyze the article title and return ONLY valid JSON:
{"relevant": true/false, "confidence": 0-100, "category": "ai|tech|business|science|other", "reason": "one sentence"}
Mark as relevant if the article covers AI tools, automation, developer technology, or tech industry news. Be strict — if unsure, mark irrelevant.
🤖 Node 5 — Summarizer System Prompt
Summarize this article in exactly 3 sentences. Be factual and specific. Include the key finding or announcement. Do not add commentary or opinions.
Article title: {{title}}
Article description: {{description}}
05 · Expand
What You Can Build With This Stack
📰
Personal News Digest
Add 10 RSS feeds across your interests. Get a daily Telegram digest of the 5 most relevant articles with AI summaries.
🔍
Competitor Monitor
Monitor competitor blogs and social feeds via RSS. Get Telegram alerts on new posts. Store everything in Supabase for trend analysis.
🤖
AI Research Assistant
Add a Telegram command node. Send /research {topic} → n8n searches feeds, summarizes findings, replies with a research brief.
📊
Content Idea Generator
AI analyzes trending articles, generates content ideas with outlines, stores them, and sends a weekly digest to Telegram.
🌐
Website Uptime Monitor
Replace RSS with HTTP Request nodes. Ping your sites every 5 minutes. If down → Telegram alert, log uptime to Supabase.
💬
AI Chatbot for Telegram
Add a Telegram Trigger node. Route to OpenRouter free models for replies. Add memory with Supabase — a free ChatGPT alternative.
06 · The Math
$0 Stack vs Paid Alternatives
| Component | $0 Stack | Paid Alternative | Our Cost | Paid Cost |
|---|
| Automation | n8n Community (self-hosted) | Zapier Professional | $0 | $29.99/mo |
| AI Models | OpenRouter free models | ChatGPT Plus + Claude Pro | $0 | $40/mo |
| Database | Supabase free tier | Airtable Pro | $0 | $20/mo |
| Notifications | Telegram Bot API | Slack Pro | $0 | $8.75/mo |
| TOTAL | $0 Stack | Paid Stack | $0/mo | $98.74/mo |
⚠️ Honest Limitations of the $0 Stack
Free AI models are slower and less capable than paid ones. Llama 3.3 70B handles classification and summarization well, but it won't match Claude Sonnet for complex reasoning. Budget $5–10/month in OpenRouter credits if you need higher accuracy.
Rate limits are real. OpenRouter free models cap at 50 requests/day. Supabase free tier pauses after 1 week of inactivity. n8n self-hosted means you manage the server. Generous for personal use, not built for hundreds of daily executions.
You need a server. n8n self-hosted requires a machine that's always on. A $4–5/month VPS solves this — count that in and the stack runs ~$5/month, still 95% cheaper than the paid alternative.
07 · FAQ
Frequently Asked Questions
Is this really $0? What's the catch?
Yes, genuinely $0. Every tool in this stack has a free tier with no time limit. The catch is limits: OpenRouter free models cap at 50 requests/day, Supabase free tier pauses after 1 week of inactivity (but keeps your data), and n8n self-hosted needs a server — a VPS runs ~$4–5/month if you go that route. The tools themselves cost nothing.
Are free AI models good enough for real work?
For classification, summarization, and drafting — yes. Llama 3.3 70B and Gemma 4 31B are surprisingly capable for the tasks in this workflow. They're not as strong as top paid models for complex reasoning, but for classifying an article or writing a 3-sentence summary, they're more than sufficient.
Do I need a VPS or can I run this on my laptop?
n8n needs to run 24/7 for scheduled workflows, so a laptop that sleeps won't work. Options: a Raspberry Pi 4 (~$35 one-time, runs silently), an old PC set to never sleep, or a $4–5/month VPS (Hetzner, DigitalOcean). The VPS is the easiest path — one Docker command and it runs forever.
Can I upgrade individual components later?
Yes — that's the point of this stack. Start with everything free, then upgrade just the piece you outgrow: switch OpenRouter to a paid model, move Supabase to Pro, or switch to n8n Cloud for managed hosting. Nothing else has to change; only the quality and limits improve.
How does this compare to a paid-model auto-blog setup?
A paid-model setup uses premium OpenRouter models for high-quality article generation and typically runs $15–25/month. This $0 stack uses only free models for lighter tasks — classification, summarization, notifications. Think of it as the entry-level version: start here, graduate later when you need production-quality content.
◆
Start Building. It Costs Nothing.
The $0 AI Automation Stack proves you don't need a budget to start automating. Five free tools, 15 minutes of setup, and a production-ready pipeline that would run $98.74/month with paid alternatives. The free tiers are generous enough for real daily workflows — not just demos.
Total Cost$0/month for the tools, ~$5/month with a VPS — 95% cheaper than paid
Setup Time15 minutes from zero to working pipeline: Docker, 3 accounts, 7 nodes
Upgrade PathStart free, upgrade one component at a time, no lock-in
Limits50 AI requests/day free, Supabase pauses after a week idle, VPS recommended for 24/7
