Let me tell you about my biggest automation failure.
I built a cold email system with 10 data sources that would automatically research prospects and generate personalized icebreakers. Sounds smart, right?
The AI kept choosing Google ratings as the icebreaker. For every single prospect. Nobody opens a cold email that says "I noticed your business has 4.2 stars on Google." It's creepy and useless.
But it got worse. For design agencies, the AI started claiming that large language models could do full 3D modeling. They absolutely cannot. I was sending emails to design professionals telling them about AI capabilities that don't exist.
I got a lot of very confused replies.
The lesson? Automation without strategy is just automated chaos. And I learned it the hard way so you don't have to.
Before we talk about what works, let's be honest about what doesn't.
Problem 1: No ICP (Ideal Customer Profile). My cold email disaster happened because I didn't define who I was targeting. I was spraying automation at everyone. Without knowing exactly who you're automating for, you'll automate the wrong things.
Problem 2: No product to sell. Automating outreach when you don't have a clear offer is like building a highway to nowhere. Fast, efficient, and completely pointless.
Problem 3: Starting with the complex stuff. Businesses hear "AI automation" and think about chatbots, predictive analytics, and autonomous agents. Then they get overwhelmed and do nothing. The best automations are boring. Gloriously, profitably boring.
Not all automation is created equal. Here's how I think about it, from simplest to most complex:
Single tasks that happen the same way every time. Copy-paste jobs. Data entry. Sending the same email with slightly different details.
Examples:
Tools: Zapier, Make, N8N (my preference because it's self-hosted and free)
Time to implement: 1-2 hours per automation
Time saved: 30 minutes to 2 hours per day, per automation
Multiple tasks chained together. An entire process, not just a single step.
Examples:
Tools: N8N, Make, or Zapier with multi-step workflows. Add AI (Claude API, OpenAI API) for the smart parts.
Time to implement: 1-2 days per workflow
Time saved: 5-15 hours per week
Workflows that make decisions. AI doesn't just execute tasks but chooses what to do based on context.
Examples:
Tools: N8N or Make with AI nodes, custom scripts, Claude/GPT API with function calling
Time to implement: 3-7 days
Time saved: 10-30 hours per week
AI that works independently on complex tasks. This is the frontier.
Examples:
This is how I work now. I use VS Code with AI coding agents that write code themselves. My work barely consists of changing code manually anymore. I can build applications 100x faster than before.
Tools: Claude Code, custom agent frameworks, AutoGen, CrewAI
Time to implement: 1-4 weeks
Time saved: Entire job functions
Stop reading frameworks. Start doing. Here are five automations that work for virtually every small business:
What: AI reads incoming emails, categorizes them (urgent, sales, support, spam), and drafts responses for common queries.
How: Set up an N8N workflow that connects to your email inbox, sends each email to Claude API for classification, and routes accordingly.
Savings: 45-90 minutes per day for anyone who handles a shared inbox.
What: After every meeting, AI generates a summary, extracts action items, and creates tasks in your project management tool.
How: Use Otter.ai or Fireflies.ai for transcription, then pipe the transcript through Claude to extract structured action items, then auto-create ClickUp/Asana/Trello tasks.
Savings: 15-30 minutes per meeting, plus dramatically better follow-through.
What: AI helps draft content, repurpose long-form content into social posts, and schedule everything.
How: Write one long piece of content (blog post, newsletter, video script). Feed it to AI with instructions to create 5 LinkedIn posts, 3 X/Twitter threads, and 2 Instagram captions. Review, edit, schedule.
Savings: 3-5 hours per week.
What: When a new customer signs up, everything happens automatically: welcome email, account setup, first-meeting scheduling, document sharing.
How: N8N/Make workflow triggered by your CRM. Each step automated with conditions (different flows for different customer types).
Savings: 1-2 hours per new customer, plus consistent experience.
What: Every Monday morning, a report appears in your inbox with last week's key metrics, comparison to targets, and AI-generated insights.
How: Connect your data sources (CRM, analytics, accounting) to an automation that pulls numbers, feeds them to AI for analysis, and formats a readable report.
Savings: 2-4 hours per week, plus you actually READ the reports because they're concise and insightful.
Deep and fast research, social media content, and coding. Those three save me hours every single day.
Research that used to take half a day now takes 10 minutes. I use scrapers (Apify is the biggest), data tools like Data4SEO, and web scrapers to get data from everywhere. Then AI analyzes it. The combination of good data and smart analysis is incredibly powerful.
Without data, you're just another person with an opinion.
Over-automating too early. Start with one workflow. Make sure it works. Then add the next one. Automating 10 things at once means 10 things break at once.
No human oversight. AI will hallucinate. It will make mistakes. Always have a human review step for anything customer-facing, especially in the early stages.
Ignoring edge cases. Your automation works great for 90% of scenarios. The other 10% will cause chaos if you haven't planned for them. Always build an "escape hatch" that routes edge cases to a human.
Not measuring results. If you can't quantify the time or money saved, you can't justify the investment. Track everything.
AI automation isn't about replacing your team. It's about freeing them from the work they hate so they can focus on the work that matters.
The businesses that thrive in 2026 aren't the ones with the most employees. They're the ones with the smartest workflows. One person with the right automations can outperform a team of ten doing everything manually.
Start small. Start today. And for the love of everything, define your ICP before you automate your outreach.
What's the best automation tool for small businesses?
N8N if you're comfortable with self-hosting (it's free and extremely powerful). Make or Zapier if you want something simpler with a visual interface. All three can connect to AI services like Claude and ChatGPT.
How much does automation typically cost?
Free to start with N8N (self-hosted) or free tiers of Make/Zapier. Expect to spend 50-200 euros per month on AI API costs (Claude, OpenAI) depending on usage volume. That's a fraction of the employee time you'll save.
Will automation replace my employees?
Not directly. Automation handles the repetitive tasks, freeing your employees for higher-value work. The result is usually a more productive team, not a smaller one. Though some roles will naturally evolve.
What's the most common automation failure?
Automating without strategy. Just like my cold email disaster: the technology worked perfectly, but the approach was wrong. Always start with "what problem are we solving?" not "what can we automate?"
Mitchell van Rijkom is the founder of AI Survivors. He's automated everything from content pipelines to application development, and helps small businesses do the same through practical, results-focused implementations.
Let me tell you about my biggest automation failure.
I built a cold email system with 10 data sources that would automatically research prospects and generate personalized icebreakers. Sounds smart, right?
The AI kept choosing Google ratings as the icebreaker. For every single prospect. Nobody opens a cold email that says "I noticed your business has 4.2 stars on Google." It's creepy and useless.
But it got worse. For design agencies, the AI started claiming that large language models could do full 3D modeling. They absolutely cannot. I was sending emails to design professionals telling them about AI capabilities that don't exist.
I got a lot of very confused replies.
The lesson? Automation without strategy is just automated chaos. And I learned it the hard way so you don't have to.
Before we talk about what works, let's be honest about what doesn't.
Problem 1: No ICP (Ideal Customer Profile). My cold email disaster happened because I didn't define who I was targeting. I was spraying automation at everyone. Without knowing exactly who you're automating for, you'll automate the wrong things.
Problem 2: No product to sell. Automating outreach when you don't have a clear offer is like building a highway to nowhere. Fast, efficient, and completely pointless.
Problem 3: Starting with the complex stuff. Businesses hear "AI automation" and think about chatbots, predictive analytics, and autonomous agents. Then they get overwhelmed and do nothing. The best automations are boring. Gloriously, profitably boring.
Not all automation is created equal. Here's how I think about it, from simplest to most complex:
Single tasks that happen the same way every time. Copy-paste jobs. Data entry. Sending the same email with slightly different details.
Examples:
Tools: Zapier, Make, N8N (my preference because it's self-hosted and free)
Time to implement: 1-2 hours per automation
Time saved: 30 minutes to 2 hours per day, per automation
Multiple tasks chained together. An entire process, not just a single step.
Examples:
Tools: N8N, Make, or Zapier with multi-step workflows. Add AI (Claude API, OpenAI API) for the smart parts.
Time to implement: 1-2 days per workflow
Time saved: 5-15 hours per week
Workflows that make decisions. AI doesn't just execute tasks but chooses what to do based on context.
Examples:
Tools: N8N or Make with AI nodes, custom scripts, Claude/GPT API with function calling
Time to implement: 3-7 days
Time saved: 10-30 hours per week
AI that works independently on complex tasks. This is the frontier.
Examples:
This is how I work now. I use VS Code with AI coding agents that write code themselves. My work barely consists of changing code manually anymore. I can build applications 100x faster than before.
Tools: Claude Code, custom agent frameworks, AutoGen, CrewAI
Time to implement: 1-4 weeks
Time saved: Entire job functions
Stop reading frameworks. Start doing. Here are five automations that work for virtually every small business:
What: AI reads incoming emails, categorizes them (urgent, sales, support, spam), and drafts responses for common queries.
How: Set up an N8N workflow that connects to your email inbox, sends each email to Claude API for classification, and routes accordingly.
Savings: 45-90 minutes per day for anyone who handles a shared inbox.
What: After every meeting, AI generates a summary, extracts action items, and creates tasks in your project management tool.
How: Use Otter.ai or Fireflies.ai for transcription, then pipe the transcript through Claude to extract structured action items, then auto-create ClickUp/Asana/Trello tasks.
Savings: 15-30 minutes per meeting, plus dramatically better follow-through.
What: AI helps draft content, repurpose long-form content into social posts, and schedule everything.
How: Write one long piece of content (blog post, newsletter, video script). Feed it to AI with instructions to create 5 LinkedIn posts, 3 X/Twitter threads, and 2 Instagram captions. Review, edit, schedule.
Savings: 3-5 hours per week.
What: When a new customer signs up, everything happens automatically: welcome email, account setup, first-meeting scheduling, document sharing.
How: N8N/Make workflow triggered by your CRM. Each step automated with conditions (different flows for different customer types).
Savings: 1-2 hours per new customer, plus consistent experience.
What: Every Monday morning, a report appears in your inbox with last week's key metrics, comparison to targets, and AI-generated insights.
How: Connect your data sources (CRM, analytics, accounting) to an automation that pulls numbers, feeds them to AI for analysis, and formats a readable report.
Savings: 2-4 hours per week, plus you actually READ the reports because they're concise and insightful.
Deep and fast research, social media content, and coding. Those three save me hours every single day.
Research that used to take half a day now takes 10 minutes. I use scrapers (Apify is the biggest), data tools like Data4SEO, and web scrapers to get data from everywhere. Then AI analyzes it. The combination of good data and smart analysis is incredibly powerful.
Without data, you're just another person with an opinion.
Over-automating too early. Start with one workflow. Make sure it works. Then add the next one. Automating 10 things at once means 10 things break at once.
No human oversight. AI will hallucinate. It will make mistakes. Always have a human review step for anything customer-facing, especially in the early stages.
Ignoring edge cases. Your automation works great for 90% of scenarios. The other 10% will cause chaos if you haven't planned for them. Always build an "escape hatch" that routes edge cases to a human.
Not measuring results. If you can't quantify the time or money saved, you can't justify the investment. Track everything.
AI automation isn't about replacing your team. It's about freeing them from the work they hate so they can focus on the work that matters.
The businesses that thrive in 2026 aren't the ones with the most employees. They're the ones with the smartest workflows. One person with the right automations can outperform a team of ten doing everything manually.
Start small. Start today. And for the love of everything, define your ICP before you automate your outreach.
What's the best automation tool for small businesses?
N8N if you're comfortable with self-hosting (it's free and extremely powerful). Make or Zapier if you want something simpler with a visual interface. All three can connect to AI services like Claude and ChatGPT.
How much does automation typically cost?
Free to start with N8N (self-hosted) or free tiers of Make/Zapier. Expect to spend 50-200 euros per month on AI API costs (Claude, OpenAI) depending on usage volume. That's a fraction of the employee time you'll save.
Will automation replace my employees?
Not directly. Automation handles the repetitive tasks, freeing your employees for higher-value work. The result is usually a more productive team, not a smaller one. Though some roles will naturally evolve.
What's the most common automation failure?
Automating without strategy. Just like my cold email disaster: the technology worked perfectly, but the approach was wrong. Always start with "what problem are we solving?" not "what can we automate?"
Mitchell van Rijkom is the founder of AI Survivors. He's automated everything from content pipelines to application development, and helps small businesses do the same through practical, results-focused implementations.
.svg){kind=icon}
