The $0 Tool Stack That Runs a 111-Workflow Business

How We Built Enterprise Automation for $20/Month

Most automation companies charge thousands for infrastructure. We run 111 workflows across three AI news agencies, build video content, deliver newsletters to thousands, and maintain four revenue tools—all on $20 per month in actual paid services.

This isn't a trick. We didn't cut corners. We just chose free tiers strategically, combined complementary tools, and knew exactly when to upgrade. The result: production capacity that would cost $2,000–5,000/month on enterprise platforms, delivered for pocket change.

Here's exactly what we use, what it costs, and why it works.

The Core Stack: What We Actually Use

Here are the nine tools running our entire business, listed by category:

Workflow Automation: n8n Cloud (Free)

Cost: $0/month

n8n Cloud is the foundation. We host 84 live workflows on their free plan. It handles everything: connecting our news source APIs, parsing content, filtering by bias, formatting for Slack, scheduling newsletter publishing, and triggering email sequences.

Why n8n over Make (formerly Integromat)? Both have free tiers, but n8n gives unlimited workflows and 1,000 executions/month. Make gives 1,000 operations/month total. At our scale (executing 10+ times daily), n8n is the only option that doesn't cap us out by day 10.

Real limitation: The free tier blocks REST API POST operations (you can't directly publish from n8n to some platforms). Workaround: we use Slack as the review gate. Content goes to Slack → human approval → manual publish. Takes 2 minutes instead of instant automation, but zero cost.

Website Hosting: Cloudflare Pages (Free)

Cost: $0/month

All 30+ pages of mewrcreate.com live on Cloudflare Pages' free tier. Unlimited bandwidth on the free plan with global CDN edge deployment. Deploys are automatic on Git push or manual via dashboard (1 minute).

Real limitation: Free plan includes unlimited bandwidth and basic features. At our traffic levels (5,000 unique visitors/month), we have plenty of headroom. The free tier is genuinely unlimited, which is ideal for bootstrapped companies scaling fast.

Email & Newsletters: Beehiiv (Free)

Cost: $0/month

Beehiiv's free plan hosts unlimited newsletters with no API posting. We publish 21 newsletters directly (manually copying content from Slack drafts). This takes 5 minutes per newsletter once a week.

Real limitation: Free plan means no automation. If we posted 7 newsletters per week, manual publishing would kill us. At our 3–4 newsletters weekly, it's sustainable. If volume exceeds 10/week, upgrade to Pro ($29/month) for API access.

Team Communication & Workflow Gating: Slack (Free)

Cost: $0/month

Slack free tier keeps the last 90 days of messages. Our n8n workflows deliver content to #run-pipeline for human approval before publishing. This is critical: Slack becomes the human-in-the-loop layer that replaces expensive workflow approvals.

Real limitation: You can't search messages older than 90 days. For us, that's fine—we archive to JSON monthly. If you need full message history forever, upgrade to Pro ($8.25/user/month), but that's overkill for most startups.

Local AI Inference: Ollama (Free)

Cost: $0/month (plus Mac Mini hardware)

We run Ollama on a Mac Mini M4 Pro (24GB RAM, £2,000 one-time cost, now running at 3 years old). It hosts five models: Llama 2, Mistral, Neural Chat, Deepseek, and OpenHermes. This powers local content drafting, research summaries, and content scoring without Claude API costs.

Real limitation: Ollama models are weaker than Claude or GPT-4. We use them for pattern-matching tasks (summarizing, categorizing, light drafting). For high-quality content that goes public, we use Claude API. For internal tools, testing, and iteration, Ollama is free and instant.

Tunnel & Bridge: Localtunnel (Free)

Cost: $0/month

Localtunnel creates a public HTTPS URL for our local Ollama instance (localhost:11434 → https://mewr-ollama.loca.lt). This bridges n8n Cloud ↔ Ollama, allowing workflows to call local models without exposing our hardware. Command: lt --port 11434 --subdomain mewr-ollama.

Real limitation: Localtunnel URLs aren't permanent (restart tunnel = new URL). We regenerate daily. It's fragile for mission-critical systems but perfect for testing and secondary inference. For production, use Cloudflare Tunnel ($0 for free plan, more reliable).

Content API & Production: Claude (Paid, ~$20/month)

Cost: ~$20/month

This is our single paid service. We use Claude API for drafting newsletters, blog posts, social content, and product descriptions. At 300,000 tokens/month input, 50,000 output, we spend roughly $18–22 depending on volume.

Why Claude over GPT-4? Claude is 3x cheaper for equivalent quality on our workflows (batch content, analysis, repurposing). GPT-4 would cost $60+/month at our usage.

Cost optimization: We use Claude for public-facing content (high quality required). For internal drafts, testing, and categorization, Ollama runs locally free. This hybrid approach keeps costs minimal while maintaining quality.

Payment & Revenue: Stripe (Pay-as-you-go)

Cost: 2.9% + $0.30 per transaction

Stripe's standard rates apply. We have 8 products with active payment links generating revenue. Since we're processing revenue (not cost), Stripe is an income line, not an expense line. Monthly payouts offset the fees completely.

Code & Backup: GitHub (Free)

Cost: $0/month

All code, configurations, and templates live in GitHub. Free plan allows unlimited public and private repos. We use it for version control on our revenue tools and automated backups.

The Complete Cost Breakdown

Here's the monthly cost in one table:

Comparison: MEWR Stack vs. Enterprise

Let's see what the same infrastructure costs on traditional platforms:

• Make (formerly Integromat) Pro: $600/month (10,000 operations) + API cost overages
• Zapier Team: $1,000+/month for 111 workflows
• Workato: $1,500–5,000/month depending on volume
• AWS Lambda + API Gateway: $0.20 per 1M requests = ~$50/month, but requires engineering time
• Claude API (high volume): Would cost $100+/month alone at enterprise scales
• Beehiiv Pro (API access): $29/month
• Cloudflare Pages Pro: $20/month (advanced features only, free tier handles most use cases)

Enterprise Total: $2,000–5,000+/month for equivalent functionality. MEWR Stack: $20/month. The leverage is 100x+.

Free Tier Limits & Real Workarounds

Every free tier has constraints. Here's what actually bites and what doesn't:

The Bites (Real Limits We Work Around)

• n8n REST API POST blocked: We gate publishing through Slack approval (+2 min per piece). Not instant, but free.
• Beehiiv no API posting: Manual copy-paste from Slack to Beehiiv (+5 min per newsletter). Scales to 10 newsletters max before hiring an editor.
• Ollama model quality: Fine for drafts, internal work, and analysis. Claude API for final published content.
• Localtunnel URL instability: Regenerate tunnel daily. Use Cloudflare Tunnel if mission-critical (it's free and more reliable).

The Non-Issues (Free Limits We Never Hit)

• n8n 1,000 executions/month: We execute 10+ times daily = 300/month. Plenty of headroom.
• Cloudflare Pages unlimited bandwidth: Our site uses ~10GB monthly. The free tier is genuinely unlimited, so no upgrade needed.
• Slack 90-day history: We archive old messages. Search is rarely needed past 2 weeks.
• GitHub free plan: No limits that matter for our use case (unlimited repos, no seat limits).

When to Upgrade: The 90-Day Rule

Free tiers are stepping stones, not permanent homes. Here's when we'd upgrade:

• Beehiiv → Pro ($29/month): If newsletters scale past 10/week. API access enables full automation.
• n8n → Pro ($50/month): If we exceed 10,000 executions/month. Current plan has 3x headroom.
• Slack → Pro ($8.25/user/month): If we hire team members or need permanent message history. Single-founder, this is optional.
• Cloudflare Pages → Pro ($20/month): Only for advanced features (image optimization, advanced analytics). Free tier is genuinely unlimited for bandwidth and executions.
• Claude API budget increase: If content volume scales 5x, budget moves to $100/month. Still cheaper than alternatives.

The pattern: upgrade when you're actively constrained, not when a limit theoretically exists. We'll never upgrade Netlify or n8n unless we scale 10x beyond current state.

The Stack in Action: Real Workflow Example

Here's how the pieces work together. Daily news aggregation for MEWR Signal (Tech/AI news):

1. n8n (free): Fetches articles from 35 RSS feeds and APIs at 5 AM daily
2. Ollama (free): Local inference summarizes each article (Mistral model)
3. Claude API (~$0.02): Scores articles by relevance and newsworthy value
4. n8n (free): Formats top 15 articles into newsletter layout
5. Slack (free): Posts formatted newsletter to #signal-daily for human review
6. Human (2 min): Reviews, adjusts if needed, approves in Slack
7. Beehiiv (free): Newsletter gets manually published to 5,000+ subscribers
8. Cloudflare Pages (free): Archive stored with CDN distribution

Total cost: ~$0.02 + labor. Total time: 7 minutes including human review. Compare to hiring a news editor at $50+/hour.

The Catch: Why Most Founders Fail With Free Tiers

Free tiers only work if you accept human-in-the-loop bottlenecks. Most founders don't. They expect:

• Full automation: Click once, everything works forever. Not possible on free tiers without engineering work.
• No learning curve: n8n, Ollama, and API integrations require setup time. You're trading money for time investment.
• Instant support: Free products have community support, not 24/7 engineers. If things break, you fix them or wait.
• No manual work: Our system requires 15–30 minutes daily for review, approval, and publishing. This is non-negotiable.

If you want zero manual work and 24/7 uptime, you need to pay. The tradeoff is real. We chose to spend time (we're solopreneurs) instead of money. Different founders will choose differently.

The Long Play: How to Scale Beyond $20/Month

As revenue grows, the stack evolves:

• Months 1–3 ($20/month): All free tiers + Claude API. Build product-market fit.
• Months 4–6 ($100/month): Upgrade Beehiiv to Pro for API automation. Scale publishing to 50+ newsletters weekly.
• Months 7–12 ($200/month): Upgrade n8n to Pro. Hire part-time editor (offset with Stripe revenue). Scale to 200+ workflows.
• Year 2+ ($500–1000/month): Migrate to managed services (Workato, Zapier, more advanced Cloudflare features) or custom AWS infrastructure. At this point, you're making enough to justify the upgrade.

The key: every upgrade should be driven by real constraints (hitting limits), not fear or comparison. Stay lean early, upgrade aggressively when revenue requires it.

By Ethan Wilmoth, MEWR Creative Enterprises LLC
Building enterprise-scale automation on a solopreneur budget. n8n, Ollama, Claude, and strategic free tiers. $20/month infrastructure that handles 111 workflows and generates revenue.