Hey everyone, I hope you are doing well.

Three topics today. All things I have been thinking about a lot lately.

In today's email we're going to talk about:

• How I am using AI to build an autonomous support system

• The Wide Event 2026 update

• Why I launched my own Shopify store

Let's go! 🔥

Learn how to grow your Shopify App with this Blueprint

I share everything from building and making $0 to growing and selling for 7-figure. You don’t need to be the best developer or the best marketer.

14-day money back guarantee

How I am using AI to build an autonomous support system

I have been thinking about customer support for a long time.

Not the "how do I answer faster" kind of thinking. More like "how do I make sure every single customer gets the right answer, every time, without me being involved."

That is a very different problem.

Most people who try to automate support do it wrong. They set up a chatbot with generic answers. The chatbot does not really understand the product. It does not know the edge cases. It does not know why things break.

And the customer feels it immediately.

So I decided to take a completely different approach.

Instead of starting with the bot, I started with the knowledge.

Here is what I did.

I exported every single support conversation from Crisp. Every question a merchant ever asked about WideBundle. Every bug report. Every complaint. Every feature request. Everything.

Then I used AI to analyze all of those conversations in batches. Not one by one. In bulk. The goal was to extract patterns.

What are the most common problems merchants run into. What are the real solutions. What are the limitations of the product that we need to be honest about. What are the things that confuse people the most.

The AI went through all of it and structured everything into a knowledge base.

But here is the interesting part.

The first results were not great.

The analysis was too shallow. It was listing symptoms, not causes. A merchant writes "my bundle is not showing up" and the AI just categorized it as a display issue. That is not useful.

So I went back and changed the approach.

I told the AI to always ask why. Why is the bundle not showing up. Is it a theme conflict. Is it a configuration mistake. Is it a known limitation. What is the root cause.

That changed everything.

The knowledge base went from a list of problems to a deep understanding of the product. Every issue now has context. Every solution has an explanation. Every limitation is documented honestly.

More importantly, I asked the AI to analyze the way we respond, the tone, the writing style, the emojis, everything. The goal was simple: ensure the AI responds exactly the way I trained my customer support experts.

I also wrote a complete document about our strategies. How do we handle conversations, what questions we ask, how do we get ask for reviews, etc.

Then I did one simple thing: Recording a video of me walking through our dashboard and commenting everything as if I was teaching someone how to use our app but without looking at the dashboard.

I make a transcript and give it to the AI as well.

And that is exactly what I needed.

Because the goal is not to build a chatbot that deflects questions. The goal is to build an AI agent that actually knows WideBundle as well as my best support person.

I am calling her Lea.

Lea is the AI support agent I am building. She will know WideBundle in depth. The common mistakes merchants make during setup. Which themes cause conflicts. When to say "this is a known limitation, here is a workaround" instead of pretending the problem does not exist.

That is the difference between a generic chatbot and a real AI support agent.

One reads from a script. The other actually understands.

I used OpenClaw to orchestrate this whole process. Export the conversations from Crisp. Analyze them with AI. Structure the results in Notion. Review. Iterate. Improve.

It is not finished yet. I am still refining the knowledge base. Still testing. Still iterating.

But the direction is clear.

The support experience most apps provide today is going to look outdated very soon. Merchants will expect instant, accurate, context-aware answers. Not template responses.

And the founders who start building that foundation now will be way ahead.

Not because the AI is perfect today. But because the knowledge base you build today is what makes the AI useful tomorrow.

That is the real asset. Not the model. Not the framework. The structured knowledge about your product and your customers.

Start there.

The Wide Event 2026

Something I have noticed over the past few months.

Every conversation I have with app founders ends the same way. We talk about AI, about growth, about strategy. And then someone says "we should meet in person to really dig into this."

Every single time.

Because there is a limit to what you can do through a screen. You can exchange ideas. You can share frameworks. You can even collaborate on projects.

But the moments that truly change your trajectory happen in person. A conversation at a dinner table. A quick chat between two talks. An introduction from someone you just met to someone who becomes your next partner.

That is exactly why I created The Wide Event.

And this year, I feel like it matters more than ever.

We are in a moment where everything is shifting. AI is changing how we build. How we support. How we sell. The founders who will navigate this well are the ones who have a strong network around them. People who understand what they are going through. People who are solving the same problems.

You cannot build that network from your desk.

The Wide Event 2026 is on May 18 at Studio Kimpton in Paris. An evening of talks, real conversations, and the kind of connections that do not happen on Twitter or Slack.

If you have been thinking about coming, this is your sign.

Tickets are available at wide-event.com. And when they sell out, that is it.

I would love to see you there.

Why I launched my own Shopify store

I recently launched a Shopify store with a friend.

And the very first thing I noticed had nothing to do with selling. It was the spam.

Every single day, agencies flood your inbox. They want to run your ads, fix your SEO, recover your abandoned carts. The volume is insane.

As an app developer, I had no idea merchants dealt with this level of noise. Now I get it.

But the real value is somewhere else entirely.

When you use your own app on a real store with real stakes, you find bugs. You spot UX issues that never showed up in testing.

There is no substitute for actually using what you build in a real scenario.

It also completely changed how I think about building Shopify apps. You start seeing things from the merchant side. The frustrations, the priorities, the tiny details that matter way more than you thought.

You become your own user. And that makes every product decision easier.

You know exactly where the pain points are. You know which feature to build next because you need it yourself.

And then there are case studies. When you own the store, you own the data. You can take screenshots, share results, build real examples without asking anyone for permission.

That alone makes it worth it. Take that step, launch something with a friend so you still have time to work on your app.

Thanks for reading!

I’ll see you in the next email, in 14 days. Until then, take care!

Mat.