Written by Andrew Sica and Roger Bales
Since we first introduced the Next Generation z/OS Experience (IBM Project Polaris) earlier this year, we’ve had a lot of interest from clients and partners who want to know more about what we’re actually building. With the Summer 2026 Tech Preview now underway, here’s a closer look at the technical approach.
Our vision for IBM Project Polaris is to reshape the z/OS user experience into something genuinely intuitive, a platform where new practitioners can build accurate mental models quickly, instead of spending years accumulating tribal knowledge before they can work confidently. That doesn’t mean changing what z/OS is. The qualities of service that make z/OS the platform of choice for the world’s most demanding workloads aren’t up for negotiation. What we’re changing is how much friction stands between a capable practitioner and the value the platform already delivers.
The goal of IBM Project Polaris is to cut the time it takes a new z/OS system programmer to reach proficiency: from 2–6 years down to 2–6 months. We are using a Jobs-To-Be-Done (JTBD) framework to focus on real tasks and outcomes that z/OS System Programmers do. This strategy is grounded in direct engagements with over 200 customer system programmers, transforming their lived experiences into a roadmap for reducing operational complexity. We are prioritizing z/OS Configuration as the first area of focus, Polaris tackles barriers to entry, accelerating the proficiency of early-tenure system programmers and enabling faster adoption of new technologies. That’s a big target, and you don’t get there with a new UI. It takes a set of technical foundations working together, each addressing a different part of why z/OS has historically been hard to learn. This post digs into those foundations: the simplification philosophy underlying all of it, and the four technical pillars we’re building in the Tech Preview: Configuration as Code, AI assistance, an API-first platform, and a modernized CLI.
Simplification: The Philosophy Behind the Work
Before getting into the technical pillars, it’s worth being precise about what we mean by simplification, because it’s easy to misread. Simplification doesn’t mean reducing what a systems programmer is capable of doing, and it doesn’t mean stripping out the depth that experienced practitioners rely on. It means making z/OS a platform where the learning curve is a reasonable on-ramp rather than a wall.
Concretely, that means three things. First, aligning z/OS with industry standards wherever it makes sense to do so: the same paradigms, tools, and conventions practitioners already know from distributed platforms. Second, removing differences that are purely historical: terminology, interfaces, and behaviors that diverged from industry norms decades ago for reasons that no longer apply, and that today serve no purpose beyond adding to the learning curve. Third, reducing complexity itself wherever possible: fewer configuration parameters to track, fewer steps to get a common task done, fewer places where a small mistake can have an outsized consequence coupled with AI to explain, guide, and in the future, enable selected automation with “human in the loop” supervision.
What simplification explicitly does not mean is erasing the differences that give z/OS its value. These are the areas where z/OS does something differently that delivers capability no other platform can match. Those are the qualities of service that make z/OS the platform of record for transaction processing at scale.
That philosophy is what connects the four technical foundations below. Configuration as Code and our CLI modernization are both about aligning with industry standards and removing historical friction. Our API-first strategy is about reducing the complexity of building and integrating with z/OS. And AI addresses, the parts of the platform that are more complicated because they deliver real value, where the right answer is to make that complexity navigable rather than to remove it.
Configuration as Code
If you’ve worked on any modern platform, you manage configuration as code, in Git, reviewed in pull requests, deployed through pipelines. On z/OS, that hasn’t been possible in a way that’s equivalent to other platforms. Configuration lives in PARMLIB and other mostly proprietary constructs that are unique to z/OS. For someone coming from a Linux or cloud background, it’s one of the first walls they hit.
The Tech Preview begins to change this. z/OS configuration is now expressed in YAML and managed through Git. You edit config files in VS Code, where schema validation catches errors before anything touches the system. Changes are versioned. History is traceable. If two systems drift apart, you can see it. The same practices your team already uses for application code now apply to z/OS configuration.
We’re starting with a focused set of PARMLIB members, with schemas developed by the z/OS component teams to make sure they’re accurate and complete. The plan is to scale out this configuration effort and expand YAML coverage across more z/OS configuration domains with each Tech Preview iteration. We’re also looking further ahead at additional configuration and policy beyond PARMLIB – watch this community for updates!
Finally, we are working in close partnership with the broader IBM Z ecosystem to ensure ‘as code’ capabilities are a full-stack solution.
AI Assistant: A z/OS Expert in the Chat
The AI experience in IBM Project Polaris is built around IBM watsonx Assistant for Z (WXA4Z), and it shows up as a chat interface embedded directly in the z/OS portal. It answers questions with relevant guidance. If you’re working through a configuration change and you’re not sure what a parameter does, you ask. It knows the system, it knows what you’re looking at, and it gives you a grounded answer instead of sending you to a 500-page manual. This ability to explain and provide guidance is a key element that we believe will accelerate time to proficiency for new-to-Z professionals.
What makes this work is the MCP (Model Context Protocol) server we’re shipping as part of the Tech Preview. The MCP server runs on z/OS and gives the assistant access to live system state data. That context is what lets the assistant give useful, specific answers rather than generic ones. If you ask, “what’s the active SMF configuration on this system?”, it can retrieve system state info to answer the question.
The MCP server is also designed to evolve quickly. New tools can be added and delivered independently of the base z/OS release cycle, so the assistant’s knowledge of the system grows continuously. Looking further ahead, we’re exploring how AI can assist with more complex tasks, such as z/OS upgrades, impact analysis, and identifying configuration drift, as we build confidence and add guardrails that make deeper AI assistance practical in production environments.
API-First: One Foundation for Everything
A key focus area of IBM Project Polaris is to build a common set of z/OS APIs to access system state information. Rather than designing for a specific mode of interaction, this will enable us to support a wide range of interfaces – it’s a design point with the future in mind.
At the foundation of this strategy is a set of native, 64-bit C APIs accessible under z/OS Unix System Services. That foundation is what makes everything built on top of it possible. New language bindings, additional MCP tools, and other modes of operation we haven’t built yet can all be layered on top of the same foundation without needing their own separate path into the system.
The goal is aggressive growth of API coverage. As that surface grows, every consumer built on top of it (the CLI, the AI assistant, etc.) gets more capable without any of those layers needing to change. That’s the point of building API-first: the foundation grows once, and everything above it can benefit from that investment.
CLI: Making z/OS Unix System Services Feel Like Home
IBM Project Polaris is additionally focused on making the UNIX command line a more natural way to manage z/OS.
The goal is straightforward: access z/OS configuration from any CLI client, log in using standard SSH, and execute commands much like you would on Linux. For people who have worked within a Linux environment, z/OS should feel like a system they can approach through familiar command-line patterns.
Today, working with z/OS often means moving between the UNIX shell, TSO, ISPF, operator commands, and other tools. That reflects the depth and history of the platform, but it also creates friction for new users. Before they can be productive, they often need to learn several different ways to interact with the same environment.
Our long-term direction is for users to be able to manage their z/OS environment from the UNIX command line. They should be able to inspect configuration, understand what is running, issue operational commands, and work with core system capabilities from a shell-based experience.
The experience we want is simple: SSH in, use the shell, and manage z/OS in a way that feels familiar to people who already know Linux and modern infrastructure tooling.
If you're looking for an overview of the broader IBM Project Polaris vision, the goals behind the Technology Preview Program, and opportunities to get involved, read Fiona King's companion blog: https://community.ibm.com/community/user/blogs/fiona-king1/2026/06/30/ibm-project-polaris-technology-preview
Built to Iterate
From the beginning, IBM Project Polaris has been informed and guided by hundreds of IBM Z professionals from customers around the world. The Tech Preview is a starting point, not a finished product. We’re releasing early because the people who work with z/OS every day are the ones best positioned to tell us where we’ve got it right and where to push further. Every area (XaC, AI, APIs, CLI) has a roadmap that extends well beyond what’s in the initial release.
The work will expand beyond configuration into the broader set of tasks that define the systems programmer role: installation and upgrade, problem determination, security and compliance management, and more. The goal is a platform where early-tenure practitioners can become productive quickly, and experienced ones can do more with less friction.
If you’re interested in joining the early access program as a sponsor user, reach out to your IBM representative or sign up for the IBM Project Polaris Sponsor User Program here: ibm.biz/PolarisSUProgram.