Why You Should NOT Start with AI (And What to Do First)

Everyone's talking about AI. Your competitor has a chatbot. Your accountant uses ChatGPT. And you're thinking: I need to do something with AI too, or I'll fall behind. So you buy an AI tool, connect it to your website, and then... it doesn't work. Or worse: it works, but it amplifies exactly the problems you already had.

This is the story of hundreds of SME entrepreneurs in the Netherlands. They jump on the AI train without the track being ready for it.

The problem: AI amplifies what's already there

This is the most underestimated risk of AI: it's an amplifier, not a fixer. If your customer service is well organized, AI makes it brilliant. But if your customer service is a mess, AI makes that mess bigger. Faster. More efficient. But still a mess.

A real-world example. An installation company implemented an AI chatbot for quote requests. Sounds logical. But their quote process consisted of: customer calls, employee writes something on a note, note goes on a pile, someone else calls the customer back. Sometimes within a day, sometimes after a week.

The AI chatbot now generated three times as many quote requests. But the sticky-note system stayed the same. Result: more leads, but also more dissatisfied customers who never got called back. The chatbot wasn't the problem. The process was the problem.

The foundation: 5 steps before you start with AI

1. Map your customer journeys

How does a customer find you? What happens after first contact? How does follow-up work? Where do people drop off?

This doesn't have to be a fancy workshop. Grab a sheet of paper and draw it out. From first contact to payment. You'll be surprised how many gaps you find. An email nobody answers. A form that goes to an inbox nobody checks. A quote that's made manually while 80% of requests are identical.

Only when you know where the gaps are, do you know where AI adds value.

2. Standardize your processes

AI works best with clear, repeatable processes. If every employee creates a quote differently, AI can't automate it. If customer data is spread across three Excel files, a notebook, and someone's head, then an AI chatbot has no source data to work with.

Start simple:

• Use a fixed format for quotes
• Store customer data in a central place (a CRM doesn't have to cost thousands -- HubSpot Free is already sufficient)
• Create checklists for recurring tasks
• Document how your team answers questions (this becomes training data for your chatbot later)

3. Digitize what's still analog

Sounds obvious, but you'd be surprised how many businesses in 2026 still work with paper order forms, handwritten calendars, or phone-based order lists. AI can't work with information that's sitting in a drawer.

This doesn't mean you need a big IT system right away. Start with the basics:

• An online calendar (Google Calendar is free)
• A shared file system (Google Drive, OneDrive)
• Digital forms instead of paper ones (Google Forms, Typeform)
• WhatsApp Business instead of personal WhatsApp for customer contact

4. Measure what you do

You can't improve what you don't measure. And you certainly can't prove AI works if you don't have a baseline.

Before deploying AI, you want to know at minimum:

• How many leads come in per week?
• What's your response time on inquiries?
• How many quotes convert to orders?
• How much time does your team spend on repetitive tasks?

These numbers are your baseline. After AI implementation, you compare against them. Without a baseline, every AI result is anecdotal: it might feel better, but you don't know for sure.

5. Set your priorities

Now that you've mapped, digitized, and measured your processes, you can see exactly where the biggest opportunities lie. Maybe it's customer service. Maybe it's lead follow-up. Maybe it's something completely different: inventory planning, staff scheduling, or invoicing.

The temptation is to deploy AI everywhere at once. Don't. Choose the process that:

• Takes the most time
• Is the most predictable (lots of repetition)
• Has the biggest impact on revenue or customer satisfaction

Start there. Learn from it. Then expand.

When are you actually ready for AI?

A simple checklist:

• You know how your key processes work (and they're documented)
• Customer data is stored digitally in a central place
• You have a baseline of your current performance
• Your team understands what's going to change (and supports it)
• You've identified a specific problem that AI can solve

If you can check off these five points, you're ready. And then AI doesn't work as a band-aid, but as a turbo on an engine that's already running.

The good news: it doesn't have to take long

The steps above might sound like a big project. They don't have to be. Most SMEs can get their foundation in order in 2-4 weeks. No six-month consultants. No half-year IT migration. Just practical, step by step.

And the beauty is: even without AI, these steps already deliver results. Better processes mean fewer errors, faster follow-up, and happier customers. AI is then the cherry on top, not the foundation.

The right order

This is the order we recommend at ZeroCode Ventures:

1. Map processes (week 1) - understand your current situation
2. Standardize and digitize (week 2-3) - make the foundation solid
3. Measure and evaluate (week 3-4) - establish your baseline
4. AI pilot (week 5-6) - start small with a specific process
5. Evaluate and scale (week 7+) - measure the difference and expand

Conclusion: AI is the answer. But not the first step.

AI is transforming how SMEs work. That's beyond question. But the businesses that get the most out of AI aren't the ones that start fastest. They're the ones that start smartest.

First the foundation. Then the technology. In that order.

Want to know if your business is ready for AI? Send us a message via WhatsApp and we'll think along with you, no strings attached. No sales pitch, just an honest conversation about where you stand and what the logical next step is.