Whenever someone asks me if we’re using #AI or #GenAI in the NETGRIF platform, I always say, “It’s on the roadmap.” But what I like to point out is that Petriflow, our #LCNC source code, is already way ahead when it comes to bridging the gap between machines and humans—more so than most low-code platforms out there.

Here’s a little experiment I did to prove the point. I wanted a new example for our academy web page, so I asked #ChatGPT to suggest 10 low-code apps I could build. I picked loan processing—something I hadn’t tried before.

Generated process after just 3 messages

Honestly, I didn’t feel like starting from scratch. So, I uploaded a few Petriflow examples I had lying around—things like handling requests, importing CSV files, and order processing. I hoped that AI could understand the context and build something around it. And guess what? It did. The prototype came with all the necessary data, roles, tasks, etc. After a few regenerations and some manual tweaking, I had a fully functional application in just 60 minutes. 🚀

So, why did this work so well?

The secret is in Petriflow. The source code is a structured XML file. When this XML file is uploaded to our running web application, it’s immediately interpreted into Java, Angular, and NoSQL databases.

Generated Loan Processing in NAB

Here you can find the Loan Processing application in the Netgrif Application Builder.

Sure, generative AI can spit out Java code or even TypeScript functions, but could it generate a complete loan processing application? That’s a different story. The reason this worked so well is that Petriflow’s structure is clear, concise, and easy for AI to understand. It didn’t need to invent anything complex—it simply pieced together what was already there in a smart way.

With just 10 example files uploaded, the AI had enough context to build the rest. And because Petriflow XML is already designed to translate into real, functioning components, the app wasn’t just a mockup. I uploaded it to our Application Builder and Application Engine and boom, it worked on the first try.

Running Loan Processing case from generated source code

TL;DR

  • Petriflow source code (structured XML) makes low-code development faster and smarter.
  • Generative AI was able to generate a fully functional loan processing app by piecing together 10 example files I provided.
  • The app ran right out of the box thanks to the way Petriflow translates XML into Java, Angular, and NoSQL databases.
  • It’s a game-changer in making application development faster, more accessible, and super efficient. But the real star of the show here is Petriflow—not AI.