2026: The Year AI Affects All IBM i Users
IBM i CTO Steve Will previews what is sure to be a big year for AI on IBM i, including the release of IBM Bob and a focus on agents
I’ve been writing IBM i blogs for over 16 years now, and there have been many significant technological advances in that time. One of my favorite parts of celebrating the IBM i 35th anniversary in 2023 was revisiting many of them in our celebratory events. Well, as the title of this post indicates, another big one is coming in 2026: AI will begin to affect IBM i users broadly.
Today, I’m going to cover the topics at a high level, and then, as you watch our announcements unfold over the year, you’ll see these get defined more specifically. Let’s get started.
Software Development – IBM Bob for IBM i Application Development
We’ve been talking about the potential of AI to help IBM i application development for almost two years now. Since spring of 2024 at COMMON POWERUp, when I highlighted the beginning of a project to apply generative AI to RPG development, we’ve described our progress in a series of blogs, articles, podcasts, webcasts and more. Over that time, we’ve expanded our goals beyond just RPG and beyond only understanding existing code.
Most of you have probably already heard about the preview of IBM Bob. It will help anyone involved in software development for IBM i. When your business invests in Bob, you’ll essentially be hiring an assistant for everyone from the CTO to the application architect, from the developer to the test engineer to the build team.
Bob will have base functionality useful for all developers on all common platforms, and will also have specific capabilities for IBM i developers, whether their code is written in RPG, COBOL, CL, SQL, Java—anything, really, that runs on IBM i. And Bob will ultimately help at every stage of the software lifecycle, and with all lifecycle tasks.
If you’re new to RPG, it will serve as a veteran RPG programmer, helping you understand, modernize and add functionality to code, even if that code was written by someone who retired a decade before you took ownership. If you’re an experienced developer who needs to architect a large new business function across your massive application, Bob will help you organize your project and guide you to the pieces that will require the most work. It will do the tedious examination of source code, data definitions and UI points, making it far more likely your new design will find all the places where your team needs to do work.
I’ve been talking about IBM Bob for a few months now, and very soon, we will have the details: when it ships, how you pay for it, what specific IBM i-related tasks we’ve validated work well, and how to accomplish those tasks.
Some members of the community have been testing Bob in their real jobs, and the reactions have been impressive. If you’re interested in an example, I suggest you follow Todd Stewart on LinkedIn. Todd works for Heartland Co-op, is a member of our COMMON Americas Advisory Council, and he has been using Bob to make his work more productive, even in its pre-GA form. He’s been posting reactions and examples. One of my favorite stories is about Bob to find a bug in record time.
Also, if you’re interested in seeing more of Bob in action for IBM i clients, we recorded a couple of Guided Tour webcasts, which are available at the COMMON Guide Tours website, and there are videos of IBM Bob on YouTube, including a collection of Bob being used on IBM i code.
Agentic Function in IBM i – MCP, Tools and More
Many of us in the technology industry have been watching the AI explosion and trying to learn the fundamental building blocks of the technology.
Agentic AI is one of those building blocks. Now, as with many technologies, agentic AI can get very complicated very fast, so let me describe at a high level the important points about agentic AI that will begin to affect IBM i and IBM i users in 2026.
Agentic AI gets its name from the fact that pieces of software called “agents” connect humans and software to AI services. One easy way to think of an agent is to imagine it’s the AI software you’re interacting with when you engage with a chatbot. When you’re interacting with an AI agent, it looks for a set of “tools” that can help it answer your question or perform action you requested. Those “tools” are really just programs or services that reside somewhere with the information or the ability to perform the action.
In the second half of 2025, we began to preview the idea that IBM i, at a system level, will have a server that can serve up tools to get information from IBM i and perform actions on IBM i. That server uses a protocol called model context protocol (MCP). Agents look for servers that can use the MCP protocol to interact with tools. Our team has made an initial version of an IBM i MCP server available, and we’ve set a goal to produce at least 500 tools in 2026. We’ve already shared several that can be used by MCP Clients to answer questions about the security of your IBM i system.
Now, IBM is certainly not the only player in the agentic AI sandbox. Several vendors who have application development and system management products are also investing in agentic support. This is great! This investment is yet another demonstration that IBM i is a modern platform. IBM and our IBM i community are adding new capabilities all the time.
Because of this investment in AI, it’s likely that a majority of the IBM i community will be using AI with their IBM i system before 2026 ends. And within a couple of years, AI will be integrated into the operations and applications of IBM i for all users.
The future is here. IBM i is ready. Let’s make 2026 another fantastic year!