I’m proud to announce that I am an AI developer. Well, I built something with AI. Let me explain.
I built an MCP integration that lets me check my invoices through chat: what I’ve received through NAFA, and what I’ve issued in my invoicing platform. Next, I want to make issuing an invoice through the conversation possible too.
I don't have a technical background. I’m not a developer, and seeing the integration work doesn’t tell me whether it’s secure or reliable.
Who can access the data? What happens if a request fails halfway through, or runs twice?
I know enough to ask some of these questions. I also know there are questions I haven’t thought to ask. Even a tool that’s “just for me” has access to real data. When we build products for other people, they’re trusting us with theirs too. I’d want a specialist involved before offering something like this to a client.
On the other hand, that can’t mean relying on someone to check thousands of AI-generated lines only at the end.
We need a process that includes their expertise throughout development: clear requirements, small changes they can review, agreed tests, and checkpoints before anything reaches real users.
Someone needs to own the decision that it’s ready, and what happens after release.
Understanding the business needs, understanding the work, and bringing in the right expertise at the right time give a project a much better chance of succeeding. Using AI myself to develop helps me have more practical conversations about both its possibilities and its risks.
I’m genuinely excited about what we can build with AI. I also think we need to recognise the limits of our own knowledge and involve people who can fill those gaps. Our customers and users deserve that care when they trust what we build with their work and their data.




