<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Daniel Barraza on Medium]]></title>
        <description><![CDATA[Stories by Daniel Barraza on Medium]]></description>
        <link>https://medium.com/@dbar915?source=rss-a4538c400d89------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*u7AYXKHzvf5m8Pur1O2JFg.png</url>
            <title>Stories by Daniel Barraza on Medium</title>
            <link>https://medium.com/@dbar915?source=rss-a4538c400d89------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Sun, 22 Nov 2026 02:02:44 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@dbar915/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Grok Bots Won Me Over. Then I Tried to Rebuild Them.]]></title>
            <link>https://medium.com/@dbar915/grok-bots-won-me-over-then-i-tried-to-rebuild-them-fe8ed5dfff40?source=rss-a4538c400d89------2</link>
            <guid isPermaLink="false">https://medium.com/p/fe8ed5dfff40</guid>
            <category><![CDATA[open-source]]></category>
            <category><![CDATA[chatgpt]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <category><![CDATA[automation]]></category>
            <category><![CDATA[ai-agent]]></category>
            <dc:creator><![CDATA[Daniel Barraza]]></dc:creator>
            <pubDate>Thu, 19 Nov 2026 04:22:50 GMT</pubDate>
            <atom:updated>2026-11-19T20:03:41.193Z</atom:updated>
            <content:encoded><![CDATA[<h4>Where I’ve landed with ChatGPT, Hermes, and the API credits already included in my plan.</h4><figure><img alt="Cover image" src="https://cdn-images-1.medium.com/proxy/1*mCLMR1pL2soG6oA7Y0WcPg.png" /></figure><p>I’ve mostly been a ChatGPT person. I wasn’t particularly interested in using Grok. Then I tried its bots and ended up liking the experience enough that I went looking for a way to recreate it somewhere else.</p><p>If you’re here for the setup, skip down to “How I wired it up.”</p><p>I could dump notes, ideas, and things I needed to handle into one chief of staff. It would route them to bots responsible for different parts of my life: personal, work, business. Those bots had their own responsibilities and context, and they could follow up with reminders, updates, or things that needed my attention.</p><p>I didn’t have to organize everything before asking for help. I could give it the rough version and let it sort out where things belonged.</p><p>Once I got used to that, I wanted something similar from whatever I used next. I already liked ChatGPT, and I wanted to get more out of the access I was paying for. Grok’s usage caps were also getting old for how I used it.</p><p>So I started looking.</p><h3>Finding an alternative was its own project</h3><p>I figured there had to be an open-source way to put something like this together.</p><p>There are plenty of tools to explore. My search took me through bot frameworks, agent platforms, coding harnesses, and model providers. I tried different approaches, including a Paperclip pilot, and kept OpenCode in my setup along the way.</p><p>The confusing part was that these tools weren’t all alternatives to the same thing. A coding harness, somewhere to run agents, and a subscription that provides model access solve different problems. They can fit together, but finding one doesn’t give you the whole experience.</p><p>And that whole experience was what I was looking for: one assistant I could talk to, specialized roles behind it, useful context that carried forward, and some initiative.</p><p>I had a hard time finding someone explaining how they’d recreated that for themselves. There was information about the individual tools, but much less that helped me work out what I’d actually need to replace the way I was using Grok Bots.</p><p>That’s why I wanted to write this. Someone else may be trying to answer the same question, and an account of what’s working for me could save them some of that digging.</p><h3>Where I’ve landed</h3><p>I’m using dot in ChatGPT as my chief of staff, with Hermes running the specialist roles on my Fedora machine.</p><p>The roles and personas came over from my Grok setup. I’d already spent time getting them to understand their responsibilities and how I wanted to work, so I brought that with me.</p><p>Hermes gives the specialists somewhere to run, with their instructions, saved context, and the tools they need.</p><figure><img alt="Architecture diagram" src="https://cdn-images-1.medium.com/proxy/1*1fKEFtYA-UQWXZMrwBFK_Q.png" /><figcaption>Dot coordinates the work. Hermes runs the specialist roles I brought over from Grok.</figcaption></figure><p>The model behind a specialist can change. <a href="https://hermes-agent.nousresearch.com/docs/user-guide/profiles">Hermes supports separate profiles</a>, so I can keep an assistant’s role and saved context while trying a different model for the work. I still have to see how well that model handles the assignment, but I don’t have to start the whole assistant over.</p><p>That flexibility is useful to me. I want access to other providers and open-weight models when they make sense, while keeping ChatGPT as the place I actually use day to day.</p><h3>The credits are a big part of this</h3><p>I’m already paying for ChatGPT. I want to get as much use out of that plan as I can before adding more expenses.</p><p>Under the current <a href="https://help.openai.com/en/articles/20001626-api-credits-benefit-for-eligible-chatgpt-subscriptions">API-credit benefit</a>, eligible Pro subscribers on the $100 tier can claim $100 in API credits per full billing period, and the $200 tier gets $200. The benefit has to be activated for a linked API organization, and unused monthly credits expire at the end of the billing cycle.</p><p>Those credits are separate from the usage included inside ChatGPT and Codex. OpenAI also excludes Codex usage through API keys from this benefit.</p><p>That left me looking for a useful way to put the API allowance to work. A lot of what I want from AI is the ongoing assistant work I’d been doing with Grok Bots. Hermes gives me a way to put those credits toward that work through the OpenAI API, while dot stays in ChatGPT.</p><p>The cost of inference has been feeling expensive to me, especially when I’m considering yet another service. I like having the option to use open-weight models, too. OpenCode Go is still in my setup as a backup source of model access, and it can <a href="https://opencode.ai/docs/console/harnesses/hermes/">supply models directly to Hermes</a>.</p><h3>So far, good enough</h3><p>I’ve only recently started using this setup. So far, I like it. I’d call it good enough, with some fine-tuning ahead.</p><p>But I also remember what getting started with Grok Bots was like.</p><p>It felt like it took forever. I kept repeating myself, correcting things, and explaining how I wanted the bots to work until it finally started to stick. The experience I liked came after that.</p><p>So I’m allowing some time for this setup to settle in, too. Bringing over the roles and personas gives me a starting point. I still need to use them here and work through what needs adjusting.</p><p>And honestly, putting it together has been a great learning experience. I’ve had to understand what each piece actually does and how it fits with the others. I’ve enjoyed that part, too.</p><p>For now, I’m sticking with it. It’s good enough to use while I figure out the rest.</p><h3>How I wired it up</h3><p>I open one chat. Dot takes the request, Hermes runs the specialist on Fedora, and the result comes back to the same conversation.</p><p><strong>What came over from Grok</strong></p><p>Before I switched, I saved the instructions for ten roles, the routines I could recover, and 654 of 697 saved facts. Some history never made it out. I kept that vault as a reference and moved the instructions and context each role actually needed into Hermes profiles for work, personal, and business.</p><p><strong>How Hermes picks a model</strong></p><p>I use the OpenAI API first so eligible work uses my credits. DeepSeek through OpenCode Go is the backup for jobs it can handle. I can also ask for DeepSeek up front if I want it for a particular job. Hermes calls the model through the provider; it doesn’t have to launch the OpenCode CLI.</p><p><strong>What I send to Fedora</strong></p><p>Dot sends the goal, the bits of context that matter, and any file the job needs. Hermes already has the role’s instructions on Fedora, so it doesn’t need my whole chat or the vault. It does the longer work there and sends back the answer or file. That keeps the handoff small. I haven’t measured how many tokens it saves.</p><p><strong>One example: the LinkedIn post</strong></p><p>Hermes drafted the first pass using the OpenAI API. I made the final edits with Dot and approved the copy. BrowserOS Neo posted it through my signed-in browser on Fedora. Dot checked the post was live and sent me the link. The model that wrote the draft never needed my LinkedIn login.</p><p><strong>What happens when a job gets stuck</strong></p><p>One job finished on Fedora, but the result never made it back to chat. We pulled the result that was already there instead of running the model again. That’s the first thing I’d test if I were doing this over: can I send one job, get the result back, and ask for a revision?</p><figure><img alt="Five steps for the LinkedIn post: request in ChatGPT, Hermes drafts on Fedora with the OpenAI API, Daniel approves the copy, BrowserOS Neo posts it, and Dot verifies the live link and returns it in chat." src="https://cdn-images-1.medium.com/max/1024/1*-uS0Eibxw9c64tLnx5587g.png" /></figure><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=fe8ed5dfff40" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[What If Security Changes Came In as Pull Requests?]]></title>
            <link>https://medium.com/modern-mainframe/what-if-security-changes-came-in-as-pull-requests-af0e7ce3d4ea?source=rss-a4538c400d89------2</link>
            <guid isPermaLink="false">https://medium.com/p/af0e7ce3d4ea</guid>
            <category><![CDATA[devops]]></category>
            <category><![CDATA[mainframe]]></category>
            <category><![CDATA[mainframe-modernization]]></category>
            <category><![CDATA[ci-cd-pipeline]]></category>
            <dc:creator><![CDATA[Daniel Barraza]]></dc:creator>
            <pubDate>Wed, 28 Oct 2026 15:37:38 GMT</pubDate>
            <atom:updated>2026-10-28T15:37:38.885Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*XoCPRfT-2np0R7CLyGIpPg.png" /><figcaption>ACF2 + CI/CD = Security as Code</figcaption></figure><p><em>A curious look at treating mainframe security like code, while keeping security professionals in control</em></p><p>I recently presented “ACF2 + CI/CD = Security-as-Code for the Mainframe” with a colleague at Broadcom’s Mainframe Technical Exchange in Plano.</p><p>While I was building the demo, I kept coming back to one question. We use Git, reviews, builds, and automated checks for application code. Why not apply some of those same practices to security rules?</p><p>I wanted to see what that could look like without asking the people responsible for security to give up ownership.</p><p>Think about a familiar access request. Somebody needs access to a data set, so they open a ticket. A security professional reviews it, asks questions, figures out the right permissions, and makes the change.</p><p>There’s a lot of expertise inside that process. Knowing the syntax is only part of it. Understanding whether the access makes sense is where experience matters.</p><p>Now imagine that same request arriving with a proposed rule change attached.</p><p>The security person can see exactly what would change, why somebody is asking for it, and which automated checks have already passed. They can ask for revisions or reject it. And the organization can limit who gets to propose changes in the first place. Maybe that’s a team lead, maybe it’s somebody else the security team trusts.</p><p>The ticket still explains the business need. The pull request shows the actual change.</p><p>That’s really the idea. Give security something more concrete to review, with evidence that follows the change. They still own the decision.</p><p>For the demo, I kept the scope small: one ACF2 dataset rule in a dedicated z/OS lab environment.</p><p>The rule source lived in Git. Jenkins ran Python checks against site policy defined in JSON. If those checks passed, the pipeline paused at a deployment approval gate. After approval, Zowe CLI uploaded the source and submitted jobs using native ACF2 utilities to compile and store it.</p><p>Then came the part I cared about most. The pipeline read the rule back from ACF2 and compared it with the Git version.</p><p>Because a successful job is useful, but I also want to know what the system actually stored.</p><p>The lab used direct Git pushes and manually started Jenkins builds. The pull-request workflow is where I’d take the idea next, with protected branches, independent review, and deployment triggers set up properly. The demo showed the deployment mechanics and an approval gate. It didn’t establish all of those production controls.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*M3zb1TmLaET9HAvSwGQyag.png" /><figcaption>The lab workflow: version the rule, check policy, approve deployment, compile and store, then read back and compare.</figcaption></figure><p>The lab workflow: version the rule, check policy, approve deployment, compile and store, then read back and compare.</p><p>We walked through four scenarios: deploy, block, detect, and recover.</p><p>First, I added an access entry matching UID(NEWGUY) to the PAYROLL rule. That changed the access rule. It didn’t create a new user account.</p><p>The change passed the checks, went through the deployment gate, compiled successfully, and matched the live rule after read-back.</p><p>The comparison checked the whole rule model supported by the demo, including permissions, mode, and wildcard deny entries. Just finding NEWGUY in the output wouldn’t have told us whether the rest of the rule was right.</p><p>Next, I proposed changing the rule from $MODE(ABORT) to $MODE(QUIET).</p><p>The site policy I configured prohibited QUIET for this rule. The guardrail failed the build before approval or any mainframe call. Jenkins skipped the deployment stages.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bw8QjtxQm4owuFID8VN4ug.png" /><figcaption>Deployment build 56 rejected QUIET mode. Approval and deployment were skipped.</figcaption></figure><p>Deployment build 56 rejected QUIET mode. Approval and deployment were skipped.</p><p>That was the blocked scenario we showed. The implementation also had checks for broad wildcard grants, but the recorded rejection was specifically the QUIET proposal.</p><p>It’s a useful distinction. Something can be valid ACF2 syntax and still violate your organization’s policy. The automated check gives you a place to catch that before deployment, as long as you’ve actually defined the policy.</p><p>Then we tried a change outside Git.</p><p>An entry matching UID(ROGUE99) was added directly through the native ACF2 workflow. Git stayed unchanged.</p><p>We manually started the same read-only detector configured for nightly execution. It captured the live rule, normalized the supported syntax, and compared it with the checked-out Git revision. The result showed the unexpected entry and its permissions.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*vmAl_IUbCjssetI2vJMLIw.png" /><figcaption>Drift build 20 identified the unexpected BACKUP UID(ROGUE99) entry and its full canonical permissions.</figcaption></figure><p>Drift build 20 identified the unexpected BACKUP UID(ROGUE99) entry and its full canonical permissions.</p><p>This is where calling Git the “source of truth” gets interesting. Git describes what you intend to have. ACF2 holds what’s actually there. Giving Git that label doesn’t magically keep the two in sync. You have to check.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*uEgcFiCYW4rQxJxHLFpDJw.png" /><figcaption>The detector compares Git and a fresh live capture. Correction goes through a separate approved deployment.</figcaption></figure><p>The detector compares Git and a fresh live capture. Correction goes through a separate approved deployment.</p><p>The detector didn’t fix anything. We reconciled the difference through a separate approved deployment of the unchanged Git version, then verified the match.</p><p>I like keeping those steps separate. A direct change might be a mistake, an emergency action, or something that should become part of the intended state. Finding it gives the team something specific to discuss.</p><p>And if the capture or comparison failed, the result was an error. We couldn’t claim a match or call it drift without a reliable comparison.</p><p>For recovery, I deployed another change, verified that it was live, and then reverted that specific commit.</p><p>The revert created a new Git revision. We deployed it through the same approval-gated process and checked the live rule again.</p><p>Rollback was the boring part, which is what you want.</p><p>Of course, compiling successfully doesn’t mean the access is right for the business. And matching Git only tells us that the supported live rule agrees with what we intended to deploy. Somebody still has to decide whether that intended change makes sense.</p><p>The evidence was another useful part of the exercise.</p><p>Successful, verified deployments produced a JSON receipt connecting the Git revision, build, rule content, deployment destination, approval fields, and comparison result. A blocked proposal kept its failure evidence and didn’t get a success receipt.</p><p>If you’ve ever pieced together a change from old emails, tickets, and console output, you can probably see the appeal.</p><p>The approval fields need some care, though. The lab recorded the declared reviewer name separately from the Jenkins account that submitted the approval. Typing somebody’s name isn’t proof that they independently approved a change. Separation of duties still needs real identity and permission controls behind it.</p><p>And these receipts contained Git, pipeline, and live-comparison evidence. Native ACFRPTRL and SMF audit collection stayed outside the demo.</p><p>I also think there’s a place for AI here, although it wasn’t part of what we demonstrated.</p><p>Picture a reviewer opening a pull request and getting a plain-English explanation next to the diff, pointing to the rule entries being changed. Or an assistant flagging access that looks broader than the ticket asked for.</p><p>It could also help somebody newer ask questions about an unfamiliar rule before discussing it with an experienced security professional. Sometimes having a place to start makes that conversation a lot easier.</p><p>I’d still want the actual rule and the business request right there. An AI explanation can miss context or misunderstand a mask. Humans still decide. The useful part would be helping them see and understand the proposal faster.</p><p>That connects to something I’ve learned working on modernization. People need a practical reason to trust a new workflow.</p><p>For a security professional, that might be a clearer request, feedback before a review, or less time digging for evidence. Those benefits have to show up in the work they do every day.</p><p>If I were trying this with a team, I’d start with one clearly bounded rule set on a test system. Bring security in from day one. Agree on what belongs in Git, write a couple of useful guardrails, and compare the intended state with the live system.</p><p>Then see what you learn.</p><p>This implementation focused on ACF2 dataset rules. Taking the idea to other rule types or security products would take its own tooling, rule model, and validation.</p><p>I’m still figuring a lot of this out. If you’ve tried something similar, I’d love to hear what worked, what broke, and what your security team needed before they trusted it.</p><p>Thanks to my colleague for joining me for the session, and to everybody who came by and asked the tough questions.</p><p><em>Opinions are my own.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=af0e7ce3d4ea" width="1" height="1" alt=""><hr><p><a href="https://medium.com/modern-mainframe/what-if-security-changes-came-in-as-pull-requests-af0e7ce3d4ea">What If Security Changes Came In as Pull Requests?</a> was originally published in <a href="https://medium.com/modern-mainframe">Modern Mainframe</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Art of Mainframe Modernization: What I Learned Selling “Change” to Green Screen Rockstars]]></title>
            <link>https://medium.com/modern-mainframe/the-art-of-mainframe-modernization-what-i-learned-selling-change-to-green-screen-rockstars-2931f8245c52?source=rss-a4538c400d89------2</link>
            <guid isPermaLink="false">https://medium.com/p/2931f8245c52</guid>
            <category><![CDATA[mainframe-modernization]]></category>
            <category><![CDATA[mainframe]]></category>
            <category><![CDATA[cobol]]></category>
            <category><![CDATA[ci-cd-pipeline]]></category>
            <category><![CDATA[devops]]></category>
            <dc:creator><![CDATA[Daniel Barraza]]></dc:creator>
            <pubDate>Mon, 15 Dec 2025 16:16:09 GMT</pubDate>
            <atom:updated>2025-12-15T16:16:09.033Z</atom:updated>
            <content:encoded><![CDATA[<p><em>In my last role, I learned transformation isn’t a science. It’s an art. And the first rule is that your “hall of fame” developers will, and should, resist.</em></p><p>Let’s get one thing straight: I did not achieve a 100% successful transformation in my last role.</p><p>But I did learn from my mistakes. Over the course of my efforts, I came to realize transformation is possible, once you have the right mindset.</p><p>Read on to see how I made mistakes, learned, and ultimately realized how to best approach large-scale transformation projects successfully.</p><p>My mission for five years as a DevOps Transformation Engineer was to modernize our zDevOps practices. This meant moving our teams from a <strong>traditional</strong>, in-house SCM to modern GitLab CI/CD pipelines.</p><p>And I failed, repeatedly.</p><p>I failed with some of our most brilliant developers. These were the “hall of fame” rockstars with 30 years of muscle memory on a 3270 “green screen” terminal. These developers were so fast and so efficient that they could run circles around a new hire using a modern IDE.</p><p>And I walked in and asked them to trade that mastery for VS Code.</p><p>This was like asking a grandmaster concert pianist to start over and learn the guitar. I learned very quickly that <strong>DevOps transformation is not a science. It’s an art.</strong></p><p>The tools, like GitLab, VS Code, and SonarCube, aren’t the solution. They’re just the paint. The <em>real</em> work is human. It’s messy. And it requires you to drop the “engineer” mindset and become part-therapist, part-salesman, and full-time relationship builder.</p><h3>My First Mistake: Leading with the “Solution”</h3><p>Like any good engineer, I started by showing my “customers” the solution. “Look at this new tech! It’s better!”</p><p>The resistance was immediate and justified.</p><ul><li>“This is slowing me down.”</li><li>“My current tools work perfectly fine.”</li><li>“This feels like a waste of my time.”</li></ul><p>They were right. My new “better” way was making their productive lives <em>worse</em>. I was asking them to “slow down to go faster later,” but I wasn’t giving them a compelling reason to trust me. I was trying to sell the <em>destination</em> (a modern pipeline) without respecting the <em>cost of the journey</em> (unlearning decades of mastery).</p><p>I had to throw the playbook out. I realized <strong>you can’t “convince” anyone.</strong> You have to continuously adapt, get creative, and find a different angle for every single team.</p><h3>The “Aha!” Moment: Stop Pushing, Start Listening</h3><p>My job wasn’t to “roll out tools.” My job was to <strong>build relationships.</strong></p><p>My entire strategy shifted. I stopped trying to convert the “rockstars,” the ones who were perfectly happy and efficient. That was a losing battle.</p><p>Instead, I went looking for a different archetype: the <strong>“Eager and Willing Champion.”</strong></p><p>In every area, in every team, there’s always <em>one</em> person who is curious, open to new ideas, or just <em>loves</em> new tech. They might not be the team lead or the “rockstar,” but they are respected.</p><p>My <em>entire</em> focus shifted to finding and empowering that one person.</p><p>They became my new customer. My goal was to make <em>them</em> successful. The single most important part of my job became:</p><ol><li>Find that one “champion” on a team.</li><li>Build a real relationship with them. Listen to <em>their</em> specific problems.</li><li>Give them a “superpower” that made their life <em>immediately</em> better.</li></ol><h3>The “Superpower” Tactic</h3><p>This is where the tech comes back in.</p><p>For many of these new champions, the “superpower” I could offer was <strong>Copilot Assist in VS Code.</strong></p><p>This wasn’t about “migrating to Git.” This was about, “Hey, I know you’re working on that REXX script. Check this out.”</p><p>The moment they saw Copilot analyze an API Swagger doc and generate code, or help them build a complex DB2 query <em>in seconds</em>, their eyes lit up.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nVOqDzQMDdEpL5lbk_t3nw.jpeg" /></figure><p>That “wow” moment did what months of my “transformation” PowerPoints couldn’t.</p><ul><li>It gave them an immediate, almost magical boost for a specific, annoying task.</li><li>It gave them a reason to <em>voluntarily</em> open VS Code.</li><li>It made <em>them</em> the “smart” one who could show their team a cool new trick.</li></ul><p>I wasn’t the “DevOps guy” forcing a change anymore. I was the “helper” who gave them a new toy.</p><p>That champion, armed with their new superpower, would then go to their team meetings and <em>naturally</em> influence their peers. A recommendation from a trusted teammate is a thousand times more effective than a mandate from me.</p><h3>What “Success” Really Looks Like</h3><p>This was a slow, patient, and deeply human process. It was about finding one champion, then another, and slowly building a <strong>coalition of the willing.</strong></p><p>Some of the “rockstars” <em>still</em> hadn’t budged by the time I left that role. And that’s okay. The organization kept running.</p><p>Here is what <em>did</em> happen:</p><ul><li><strong>We built momentum.</strong> Adoption of modern tools like VS Code began to spread organically, team by team.</li><li><strong>We improved collaboration.</strong> For the teams that <em>did</em> migrate, their ability to collaborate with offshore resources and streamline code reviews with merge requests improved dramatically.</li><li><strong>We became “talent-ready.”</strong> We built a modern platform that could attract new developers who <em>expect</em> these tools, while <em>also</em> proving that the mainframe itself (running Python on Z, integrating with Zowe) is a powerful, modern environment.</li></ul><p>The biggest lesson I learned is that technology is easy. People are hard.</p><p>The “art” of transformation isn’t about having the best tools. It’s about having the most empathy. It’s about respecting the mastery of the people you’re trying to “help,” and having the patience to build one relationship at a time.</p><p>And ultimately, I learned that a “full migration” mandate might be the wrong goal entirely. Our organization wanted one toolchain, but a <strong>hybrid approach</strong>, which was not considered, is often the smarter one.</p><p>The real goal shouldn’t be 100% conversion; it should be 100% productivity. And that means giving each engineer the best tool for <em>their</em> specific job, even if that means supporting two ways of working.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2931f8245c52" width="1" height="1" alt=""><hr><p><a href="https://medium.com/modern-mainframe/the-art-of-mainframe-modernization-what-i-learned-selling-change-to-green-screen-rockstars-2931f8245c52">The Art of Mainframe Modernization: What I Learned Selling “Change” to Green Screen Rockstars</a> was originally published in <a href="https://medium.com/modern-mainframe">Modern Mainframe</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Part II: What Helped Me Survive OCD. The Tools, The Work, and Why I Still Believe in Full Recovery.]]></title>
            <link>https://medium.com/@dbar915/part-ii-what-helped-me-survive-ocd-the-tools-the-work-and-why-i-still-believe-in-full-recovery-c88bb54aaa64?source=rss-a4538c400d89------2</link>
            <guid isPermaLink="false">https://medium.com/p/c88bb54aaa64</guid>
            <category><![CDATA[depression]]></category>
            <category><![CDATA[ocd]]></category>
            <category><![CDATA[mental-health]]></category>
            <category><![CDATA[anxiety]]></category>
            <category><![CDATA[general-anxiety-disorders]]></category>
            <dc:creator><![CDATA[Daniel Barraza]]></dc:creator>
            <pubDate>Wed, 03 Dec 2025 18:10:27 GMT</pubDate>
            <atom:updated>2025-12-03T18:10:27.609Z</atom:updated>
            <content:encoded><![CDATA[<h3>If you read Part I, you know I’m still in the storm.</h3><p>But here’s what I want to make absolutely clear:</p><p><strong>I believe full recovery is possible.</strong></p><p>Not overnight. Not by luck. Not by finding “the one thing that finally fixes everything.”</p><p>Recovery, real recovery, is more like rehabbing a long-term physical injury.</p><p>When your back goes out, you don’t pray for it to magically pop back into place. You change how you sit. How you walk. How you lift. You strengthen weak muscles, stretch tight ones, fix your posture, and show up to physical therapy even when it hurts.</p><p>Some days you feel worse before you feel better. Some days you swear nothing is changing. But consistency shifts the trajectory.</p><p>Mental health works the same way.</p><p><strong>Anxiety.</strong> <strong>Trauma.</strong> <strong>OCD.</strong></p><p>These aren’t character flaws. They’re injuries to the mind and nervous system. And healing requires maintenance, adjustment, daily work, and patience.</p><p>The tools I’m about to share aren’t quick fixes. They’re not silver bullets. They are the things that slowly, painfully, steadily move you forward. They are the mental equivalent of stretching, strengthening, retraining, and rebuilding.</p><p>They are the reasons I believe recovery isn’t just possible. It’s probable if you commit to the long-term process.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*7i646QoioYAnz1ZYgClIBg.png" /></figure><h3>1. The Brain Energy Model: Supporting the Hardware</h3><p>Just like physical rehab starts with stabilizing the injured area, mental rehab starts with stabilizing the brain itself.</p><p>The <strong>Brain Energy</strong> model (Chris Palmer) taught me that mental illness isn’t just emotional. It’s metabolic.</p><p>Keto, fasting, lowering inflammation, better sleep. These changes stabilize the hardware so the software can reboot.</p><p>I don’t think keto “cures” OCD. But I do believe it gives the brain the foundation it needs to repair. It’s slow. It’s demanding. But so is healing any injury.</p><h3>2. Acceptance &amp; Commitment Therapy: Fixing the Movement Patterns</h3><p>If OCD is a movement disorder of the mind, a pattern of reacting, resisting, analyzing, then ACT is the physical therapy that retrains those patterns.</p><p>ACT teaches:</p><ul><li>Stop fighting thoughts.</li><li>Stop arguing with fear.</li><li>Stop trying to outthink your way out of a thinking problem.</li><li>Move toward your values even when fear screams at you.</li></ul><p>At first, it feels wrong. It feels weak. It feels like surrender. But it’s actually the mental equivalent of rebuilding your gait after an injury. Painful, awkward, necessary.</p><h3>3. Emotional Teachers Who Don’t Sugarcoat</h3><p>These people didn’t heal me, but they helped me understand the injury:</p><ul><li><strong>Drew Linsalata (The Anxious Truth):</strong> The mechanic of anxiety. He explains exactly why you feel the way you do and why running from it always makes it worse.</li><li><strong>Shannon Shy:</strong> A Marine’s approach to OCD. Blunt talk on how to stop “feeding the beast” and shift your attention elsewhere.</li><li><strong>Paul David (At Last a Life):</strong> Teaches the art of “allowing.” How to stop adding second fear to first fear so the nervous system can naturally cool down.</li><li><strong>Mark Freeman:</strong> Focuses on mental fitness over mental illness. He treats compulsions like bad habits you have to train yourself out of.</li><li><strong>anxietyMD (Dr. Russell):</strong> Demystifies the physical symptoms. He explains that your racing heart and dizziness are biology, not impending doom.</li><li><strong>Schwartz (Brain Lock):</strong> The classic manual on how to manually shift gears when your brain gets stuck in a loop.</li><li><strong>Claire Weekes:</strong> The original voice of acceptance. She taught the world how to “float” past fear instead of fighting it.</li><li><strong>“Nothing Works: A Letter to Myself”:</strong> A raw reminder from the online anxiety community that sometimes the only way out is to stop trying so desperately to escape.</li></ul><p>They didn’t offer magic. They offered principles. The same way a PT offers exercises you have to commit to day after day.</p><h3>4. Trauma + Nervous System Healing: Strengthening the Deep Muscles</h3><p>You can strengthen your back all day, but if your diaphragm is tight, your psoas is locked, your core is weak, the pain won’t go away.</p><p>The nervous system works the same way.</p><p>Books like <em>The Body Keeps the Score</em> and <em>When the Body Says No</em> taught me that: <strong>You can’t think your way out of a dysregulated body.</strong></p><p>Trauma release exercises, yoga, stretching, breathwork. These aren’t “wellness trends.” They’re the deep stabilizing muscles of the mind.</p><h3>5. Meditation &amp; Mindfulness: Rebuilding Stability</h3><p>Meditation isn’t about becoming calm. It’s about becoming skilled at not reacting. It’s the mental version of holding a plank. Shaky at first, painful, awkward, but over time, it becomes your foundation.</p><h3>6. Working Out: The One Proven Antidepressant</h3><p>Movement is medicine. Weights, running, yoga, CrossFit. This is the part of rehab you can’t skip.</p><p>If the mind is loud, move the body. If the body is strong, the mind has something to anchor to.</p><h3>7. Entheogens: Opening a Window</h3><p>Not a cure. Not a shortcut. But sometimes these experiences show you what a healthier mind feels like for a moment. Just enough to remind you that the state still exists.</p><p>It’s not the healing. It’s the glimpse.</p><h3>8. The Hardest Truth: Your Personality Can Injure You</h3><p>This took me years to understand. My superpower is also my vulnerability.</p><ul><li>My intensity.</li><li>My drive.</li><li>My hyper-awareness.</li><li>My logic.</li><li>My problem-solving.</li><li>My responsibility.</li><li>My sensitivity.</li><li>My discipline.</li></ul><p>These traits made me successful. But when the brain is injured, those same traits mutate.</p><p><strong>Strengths</strong> become <em>compulsions</em>.<br><strong>Drive </strong>becomes <em>torture</em>.<br><strong>Intelligence </strong>becomes a <em>trap</em>.<br><strong>Vigilance </strong>becomes <em>fear</em>.<br><strong>Discipline</strong> becomes <em>self-punishment</em>.</p><p>It’s like someone rehabbing a knee injury trying to sprint through the pain. The very instinct that once made you strong now re-injures you.</p><p>To heal, I had to stop using the part of me that got me into the loop.</p><h3>9. Healing Is Maintenance</h3><p>Mental health is not a one-and-done fix. It’s a lifelong practice. Just like strength training, mobility work, eating right, and caring for your body.</p><p>There are days where you feel worse. There are setbacks. There are flare-ups. This doesn’t mean you’re broken. It means you’re human.</p><p>Recovery isn’t perfection. <strong>It’s resilience.</strong></p><p>And with consistent work, the storms get quieter, the windows of peace get longer, and the setbacks lose their power.</p><h3>If You’re Reading This and You’re Struggling</h3><p>I want you to hear this from someone still doing the work every day:</p><p><strong>You can get better.</strong> <strong>You can heal.</strong> <strong>You can recover fully.</strong></p><p>Not by accident, but through daily, intentional rehabilitation of the mind and body.</p><p>Just like a physical injury, healing feels slow until suddenly you look back and realize you’ve climbed a mountain without noticing.</p><p>Keep going. You’re doing more than you know.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=c88bb54aaa64" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Part I — When the Mind Breaks: My Story of OCD]]></title>
            <link>https://medium.com/@dbar915/part-i-when-the-mind-breaks-my-story-of-ocd-10a0aea37242?source=rss-a4538c400d89------2</link>
            <guid isPermaLink="false">https://medium.com/p/10a0aea37242</guid>
            <category><![CDATA[anxiety]]></category>
            <category><![CDATA[depression]]></category>
            <category><![CDATA[ocd]]></category>
            <category><![CDATA[mental-health]]></category>
            <dc:creator><![CDATA[Daniel Barraza]]></dc:creator>
            <pubDate>Wed, 03 Dec 2025 17:21:56 GMT</pubDate>
            <atom:updated>2025-12-03T17:21:56.575Z</atom:updated>
            <content:encoded><![CDATA[<h3>Part I — When the Mind Breaks: My Story of OCD</h3><p>Most people know the outer version of me, the achiever, the protector, the guy who steps up, stays steady, and handles business.</p><p>For most of my life, that’s the only version I let people see.</p><p>What almost no one knows is that, for over a decade, I’ve been fighting a silent war inside my own mind.</p><p><strong>Anxiety.</strong> <strong>OCD.</strong> <strong>Depression.</strong></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yEipmPglvCiaH5GNb-Jytw.png" /></figure><p>The kind of mental storm that doesn’t show up on your face but leaves you exhausted in your soul.</p><p>Some days, the weight is manageable. Some days, it crushes me before I even open my eyes.</p><p>I used to believe strength meant staying in control, outthinking every problem, outperforming every weakness. I lived like that for years, pushing, grinding, pretending everything was fine because that’s what I thought “being a man” meant.</p><p>But there comes a point where the mind shakes you hard enough that the old definition of strength just falls apart.</p><p>That’s where my real story begins.</p><p>When OCD hit me at full force, it wasn’t a spiral. It was a collapse. A constant internal alarm. A nervous system in overdrive. A voice that never shuts up. Thoughts that don’t feel like thoughts. They feel like threats.</p><p>And the worst part? This storm didn’t hit during a low point in life. It hit when things were beautiful. A growing family, a career leveling up, a future that looked stable on the outside.</p><p>It’s a strange kind of pain to hold joy in one hand and fear in the other.</p><p>There were afternoons I was leading teams, solving complex engineering problems… while internally fighting a battle so intense I had to remind myself to breathe.</p><p>This disorder humbled me. It stripped away the illusion that I could “power through.” It forced me to face pieces of myself I never wanted to see.</p><p>And in the darkest moments, when language failed, I turned to music. I created a song just to get the emotions out of my body. Not for the world. For survival.</p><p>If you want a window into those nights, here’s the track I made:</p><blockquote><strong><em>👉 My song:</em></strong><em> </em><a href="https://youtu.be/NRGkUDzIiME?si=IvtWqutmJ-CoTozj"><em>Click to Listen on YouTube</em></a></blockquote><blockquote>(Headphones recommended. It’s a feeling more than a story.)</blockquote><p>I’m not healed. I’m not on the other side. I’m learning, every day, how to walk with a brain that doesn’t always cooperate and a heart that refuses to quit.</p><p>Some days I win. Many days I don’t. But I’m still here. Still fighting. Still showing up for my wife, my kids, my faith, my purpose.</p><p>I’m sharing this because someone out there is silently drowning the way I was, the way I still do sometimes, believing they’re alone.</p><p>You’re not. There are more of us than you think. Behind strong faces are battles nobody talks about.</p><p>This isn’t a success story. This isn’t a motivational speech.</p><p><strong>This is a survival story, still being written.</strong></p><p>And if you’re fighting too, here’s what I’ve learned:</p><ul><li>You can be weak and strong at the same time.</li><li>You can be scared and still show up.</li><li>You can be broken and still be valuable.</li><li>You can lose the day and still win the life.</li></ul><p>I’m 34 now. A father of two. A husband. A man still learning himself. A man still finding peace in the storm.</p><p>This is my truth. My reality. My battle. And my reminder, to you and to myself, that surviving another day is also a kind of victory.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=10a0aea37242" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>