Claude Code + n8n Automation Agency: Solo Operator Playbook

I stumbled on this story while researching how indie developers actually make money with AI, and the numbers genuinely surprised me.

Core Insight

If you're searching "can one person build a profitable AI business in 2026," the short answer is: yes, solo founders are hitting $10K-$50K MRR with $87/month tool stacks, zero employees, and AI agents doing 80% of the work. This case study breaks down exactly how Marcus Chen, a former marketing operations manager, built AIflow Agency from his apartment in Austin, reaching $15K MRR in 4 months serving 12 B2B clients with AI-powered workflow automation.

Project Background

Marcus Chen spent 7 years in marketing operations, manually stitching together CRMs, email platforms, and analytics dashboards for mid-sized companies. In January 2026, he was laid off during a restructuring. Instead of job hunting, he decided to productize his workflow automation expertise.

The bet: Companies pay $1,500-$3,000/month for custom automation workflows, but building them used to require 40-80 hours of development each. If AI coding agents could cut build time to 4-8 hours per client, this could be a one-person business.

Timeline:

• Week 1-2: Built internal automation templates using Claude Code + n8n, tested on 3 free beta clients
• Week 3-4: Launched landing page (Cursor + Vercel), got first 2 paying clients at $1,000/mo each
• Month 2: Scaled to 6 clients, revenue hit $6K/mo. Automated onboarding with Make + ChatGPT
• Month 3: Added retainer model ($1,500-$2,500/mo per client), 9 clients, $12K MRR
• Month 4: 12 clients, $15K MRR. Marcus works 25 hours/week. AI agents handle monitoring, alerts, and 70% of client communication drafts

Tool Stack

Revenue Sources

• Revenue 1: Workflow Automation Retainers , $10,500/mo (70%) 7 clients on $1,500/mo retainers for ongoing automation management, monitoring, and monthly optimizations. Each client has 3-8 active n8n workflows handling tasks like lead routing, invoice generation, and Slack notifications.

• Revenue 2: Custom Build Projects , $3,500/mo (23%) One-time automation builds at $1,500-$3,500 per project. Average 1-2 new builds per month. These are complex workflows (multi-step AI pipelines, legacy system integrations) that take 4-8 hours each with Claude Code assistance.

• Revenue 3: Workflow Templates Marketplace , $1,000/mo (7%) Selling pre-built n8n workflow templates on Gumroad. Top seller: "AI Lead Scoring + Slack Alert" template at $49. Passive income growing ~15% month over month.

Replicable Steps

Step 1: Pick Your Automation Niche, Solve a Boring, Expensive Problem

Marcus didn't chase the hot AI wrapper trend. He targeted a boring problem he knew intimately: companies waste 15-25 hours/week on manual data transfer between SaaS tools. His first 3 clients came from his LinkedIn network, former colleagues who knew he could fix their Zapier spaghetti.

Actionable: Make a list of 10 SaaS tools you've personally used in your career. Identify the top 3 integration pain points between them. That's your niche.

Step 2: Use Claude Code to Build Your First 3 Workflows (Week 1)

This is where AI flips the economics. A custom HubSpot-to-Slack lead notification workflow used to take 6-8 hours to code and test. Marcus describes his Claude Code workflow:

Prompt: "Build an n8n workflow JSON that:
1. Polls HubSpot API every 5 minutes for new deals
2. Enriches with Clearbit company data
3. Scores leads using this criteria: [criteria]
4. Sends formatted Slack message to #sales channel
5. Logs everything to Supabase"

Claude Code generated the complete n8n workflow JSON in under 2 minutes. Marcus spent 30 minutes testing and tweaking edge cases , a 16x speed improvement over manual coding.

Key insight: Claude Code isn't just writing code, it understands n8n's node-based architecture, API authentication patterns, and error handling. When a workflow breaks, paste the error into Claude Code and it suggests fixes with context.

Step 3: Layer AI Agents for Client Operations (Month 2-3)

Once 6 clients were onboarded, Marcus faced the solo-founder bottleneck: client communication. He built a lightweight AI agent system:

• Onboarding Agent (OpenAI GPT-4o + Make): Sends welcome email, schedules kickoff call via Calendly, creates client folder in Google Drive, populates starter n8n workflows
• Monitoring Agent (Custom Python + Claude Code): Checks all client workflows every 15 minutes, alerts Marcus via Slack if any workflow fails, auto-generates weekly performance reports
• Support Triage Agent (GPT-4o + n8n webhook): Client emails go to a shared inbox → AI classifies urgency and drafts response → Marcus reviews and sends

Result: Client communication time dropped from 15 hours/week to 4 hours/week. The AI drafts are 80% accurate; Marcus edits tone and adds context.

Step 4: Build a Client-Facing Dashboard for Transparency (Month 3)

Clients kept asking "is my automation running?" Marcus used Cursor to build a simple client portal (Next.js + Vercel + Supabase) showing:

• Real-time workflow status (green/yellow/red)
• Monthly automation volume (tasks processed)
• Time saved estimate (hours/month)
• AI-generated monthly summary

This single feature reduced support tickets by 60% and became a key selling point in sales calls. Clients could see exactly what they were paying for.

Cost to build: 3 hours with Cursor. Zero additional hosting cost (Vercel free tier).

Step 5: Productize Your Templates for Passive Income (Month 4+)

After building 40+ custom workflows, Marcus identified 8 patterns that 80% of clients needed. He generalized them into templates and listed them on Gumroad at $29-$79 each.

Top 3 templates:

1. AI Lead Scoring + CRM Sync ($49) , 120+ sales
2. Invoice-to-Slack Payment Alert ($29) , 85+ sales
3. Multi-Channel Customer Feedback Aggregator ($79) , 60+ sales

This added ~$1,000/mo in nearly passive income. Marcus updates templates monthly using Claude Code to add new features based on buyer feedback.

Risks & Pitfalls

• Pitfall 1: n8n self-hosting requires maintenance. Railway makes it easy, but you're responsible for updates, backups, and security patches. Budget 2-4 hours/month for DevOps. Marcus eventually hired a freelancer on Upwork ($200/mo) to handle this.

• Pitfall 2: AI-generated workflows can have subtle bugs. Claude Code produces working code fast, but edge cases (rate limiting, API downtime, data format changes) still need human review. Marcus learned to always add error-handling nodes and monitoring alerts.

• Pitfall 3: Client dependence on you. If you're the only person who understands the workflows, you can never truly disconnect. Marcus mitigated this by creating Loom video documentation for every client workflow and training a VA ($15/hr, 5 hrs/week) to handle Tier-1 support.

• Pitfall 4: Pricing creep. Early clients got $500/mo deals. Raising prices on existing clients is hard. Marcus now quotes $1,500/mo minimum for new clients and grandfathers early adopters at their original rate.

• Pitfall 5: API costs can surprise you. One client's workflow accidentally triggered 50K OpenAI API calls in a weekend (a loop bug). Cost: $78. Marcus now enforces rate limits and spend caps on all client workflows.

Marcus's advice: "Don't try to build a 'platform.' Build workflows that solve specific, painful problems for specific people. Let Claude Code do the heavy lifting, but you need to be the domain expert who knows what to build. The AI is a force multiplier, not a replacement for industry knowledge."

Claude Code, n8n, Cursor, OpenAI, Make, and other tools mentioned in this case are automatically detected by the platform for tool cards and hover previews.