Blurred background fast…

COBOL just wiped $30 billion off IBM's market cap in a single day.

Can we agree that things are moving fast?

Super fast. Blurred background fast.

The AI evolution and growth in the past weeks has been nothing short of super important to keep an eye on. I am both working and following several things in the development world because this is a revolution that is not a bubble and not the new thing that will someday be old.

It’s different today.

The software and data development world is now different. Coding experience is growing irrelevant. It is not changed as much as things ended. Not everywhere all at once, but in some ways, 2025 is closer to the 1960s than 2026.

What matters now is that if you are a coder that you start taking AI seriously. Really seriously.

Two Things:

I have two things that are driving a narrative in my mind about the future of software and data development.

  1. I am invested and listening. I am also concerned. There are individuals that are 100% skin in the game who are writing, talking, and proving what AI is for technology and the future of applications and everything we use daily. Listen and read. Then react. Do not live in a silo or under a rock, invest now, learn now, and be extremely familiar with the AI toolset as it is today because your way of making a living depends on it.

  2. AI is not perfect, but when the two largest AI resources are writing themselves, millions of lines of code and files, we better take notice. I am a voracious developer. I am fast, resourceful, can write in several language technologies, and known for getting things done. What AI is now makes me realize that I better count on it to help me and work for me than for me to fight the good fight and lose.

Recent Holy Crap Moments:

The Shumer Blog. I keep referring to it, take my advice and read it or open it in your phone and have the browser read it to you. Take it seriously. https://shumer.dev/something-big-is-happening

Open Claw. You should investigate this as this is a map point on the timeline of AI. This is worth your time. https://youtu.be/YFjfBk8HI5o?si=WOueU8_ek2A0lbL9

Anthropic and Cobol. The blog post was from Anthropic (the company) rather than specifically from Dario Amodei personally, and it was about using Claude Code for COBOL modernization. It made waves because it actually wiped ~$30 billion off IBM's market value. https://claude.com/blog/how-ai-helps-break-cost-barrier-cobol-modernization

Here is the point.

Most of my community is the IBMi RPG community. It as a whole is amazing. We have the same problems with moving on into the future as COBOL and the IBM mainframe. With the recent anticipation of BOB and the growing adoption of VS Code for i, things are starting to turn towards the better for us developers. But with that said, we are facing.

  • Billions of lines of code running in nearly every business domain for mission critical applications. Most IBMi code bases are anywhere from a million to several million lines of code. Most of the code has not seen the light of an editor in decades. We still use SEU/PDM, we still write programs for the 5250 interface, and although we use modern tools, we create retro antique applicatons. Usually its not our choice but we still do.

  • The talent pool of experienced programmers on the IBMi platform is shrinking, and the knowledge is not transferring very well. It is a combination of many factors that includes at the top that interest in the platform is hard to cultivate, retiring senior developers are seen as irrelevant to younger developers, and most systems have not be kept in current working order. Can I say System 36 Evoke.

  • Modernization efforts are difficult, expensive, and understanding the applications, dependencies, and nuances built into code for multiple decades, is extremely difficult as an understatement. Multiple generations of developers and a consistant style methodology has never been part of the RPG world. It exists on every other modern coding platform, but not across this platform.

  • In reality, we also face mergers and aquisitions, new generations of leaders, and the constant image that this system is old, outdated, and needs replaced. That is not true, but it is projected up into the leadership suite. Although some big companies that we all know of use the platform, the rest of us are just a management or private equity decision from relevance to obscurity.

The tension is real

The tension between those who have a skeptical view and those who see the AI future is real. It is understandable, and predictable. The tension will be found everywhere and I am sure it will certainly be found in the development teams in companies of all sizes. I am sure it will be found in the readers of this blog post.

I am not trying to sell a way of thinking or promoting AI. It is here and it is coming fast to disrupt everything. All I ask is that you take a pragmatic look at it, read and listen, and discover more so you can position yourself in the best way for the future.

What To Do Monday Morning

Stop waiting for permission. Here are three things I believe you can do this week that cost nothing but time. Take this seriously.

1. Point Claude at a piece of your RPG code today. You don't need Claude Code, a license, or a modernization project. Open claude.ai, paste in a self-contained RPG program — something with business logic you know well — and ask it to explain what the code does, map its dependencies, and flag any risks. Do it with something you already understand, so you can evaluate how well it performs. This isn't a commitment to anything. It's reconnaissance mission. You need to know what this tool can and can't do on your code before someone above you asks.

Another tip is to have it document that code into a markdown file with the key inputs, outputs, core logic, business rules, and other important things. Save the markdown and use a markdown editor plug in for VS Code and view it. I do this often and is a key tool in my arsenal.

2. Read Anthropic's Code Modernization Playbook. The COBOL blog is the headline, but the Playbook is where the methodology lives — how to run the analysis phase, how to sequence modernization work, how to prompt AI tools for code migration. It was written for COBOL, but the phases map directly to RPG: discovery, dependency mapping, risk assessment, incremental migration. Read it as if every mention of COBOL says RPG. It's a short read and it will give you a framework.

This is my current reading. I mentioned in the blog about having your phone read it to you. I get tired eyes, and find that listening helps me think as the words come across. I am taking this seriously.

3. Follow what's happening in the IBMi community right now. This conversation is already underway. IT Jungle has been covering AI and RPG closely — Fresche Solutions' X-Analysis tool already runs Claude under the hood to analyze RPG codebases. IBM's own Project Bob (the successor to watsonx Code Assistant for i) is in tech preview and covers RPG, CL, and SQL. The community is not standing still, and you shouldn't either. Subscribe to IT Jungle, join the COMMON community, and pay attention to what the vendors you already work with are quietly building on top of AI.

Claude helped me with #3. I won’t lie. It found the articles I needed. Stay vigilant.

The bottom line: you don't need to have answers yet. You need to have hands-on experience. The developers who will matter in two years are the ones who started experimenting this week.

Things are moving fast. Super fast.

Additional Resources:

  • Anthropic's Code Modernization Playbook — a companion to the COBOL blog with step-by-step guidance: https://resources.anthropic.com/code-modernization-playbook

  • Claude Code itself — link readers directly to try it: https://claude.ai/code

  • IBM's response — the story is richer because IBM pushed back saying "decades of hardware-software integration cannot be replicated by moving code." Citing that tension adds credibility and nuance.

  • Podcasts - You can find them in any podcast app. Some are more relevant than others depending on the episode and guests.

    • Lex Fridman

    • The Changelog

    • Tech Brew Ride Home (recommended by me)

    • Free Code Camp Podcast

    • The Stack Overflow Podcast

Next
Next

Can we talk?