DevOps Paradox

What is DevOps? We will attempt to answer this and many more questions.

DOP 350: Context Is the New Bottleneck, Not Code 05/13/2026

DOP 350: Context Is the New Bottleneck, Not Code #350: The bottleneck used to be writing the code. Now it is feeding the agent enough context to write the right code. That is Patrick Debois' argument, and given that Patrick coined the term DevOps, it is worth paying attention when he says the discipline is shifting again. The model does not matter. The IDE does not matter. What matters is whether your team can capture the way you actually work and hand it to an agent that does not know any of it. The promise was that AI would let us ship without writing specs. The reality is the opposite. If you want decent output, you need richer specs, more docs, and a way to feed the agent what is unique about your team and your codebase. Viktor admits he stopped writing specs himself. He talks to the agent until he is satisfied, then says write it down. The work did not go away. It moved. A second agent that validates your work tends to take the original spec too seriously and miss what is not there. The interesting validation is not whether the code matches the spec. It is whether the spec matches reality. Patrick's response is harness engineering -- combining verifier agents with deterministic tooling like linters and tests, and mining conversation logs for the moments a user says this is wrong so the missing context can be saved and reused. Memory, hooks, skills, registries -- all just delivery mechanisms for the same underlying thing. Patrick's number one piece of advice if you are starting today is brutal in its simplicity. When the agent does the wrong thing, write it down in your AGENTS.md or claude.md. Do not just re-prompt and move on. Build the context file. That is the new job. Code moved to context. Context, eventually, moves to knowledge -- the way your organization actually works, captured somewhere an agent can use it. Whoever owns that layer wins. The model does not. Patrick's contact information: LinkedIn: X: YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/41235155

DOP 349: Shadow AI Is Going to Be a Thousand Times Worse Than Shadow IT 05/06/2026

DOP 349: Shadow AI Is Going to Be a Thousand Times Worse Than Shadow IT #349: Every platform you already own is about to have AI baked into it. Not next year. This year. That is Ben Wilcox's blunt prediction, and Ben is the CTO and CISO at ProArch, so when he says shadow AI is going to make shadow IT look quaint, it is worth slowing down to figure out what that actually means. The data leaves your stack through tools you already paid for, through features the vendor shipped without asking, through copilot agents nobody filed a ticket for. Here is the uncomfortable part. This is not a new problem. It is the exact same retroactive-security failure pattern that broke DevSecOps, just with higher stakes and a faster clock. A pen test done six months ago is already obsolete because the app added AI in the meantime. Models get deprecated on seven-month windows while frameworks still get years of support. The whole "we will deal with it at the end" approach that worked badly for cloud and worked worse for containers is going to be catastrophic for AI. The fix is older than the problem. Landing zones. Well-architected frameworks. A storage account that already has the right policy. An API gateway already in front of the API. The developer should not be picking from twenty checkboxes to figure out which combination is secure -- that decision should already be made before the ticket lands. Stop forcing developers onto the security team. Stop running security reviews while the head developer sweats through his shirt right before release. Build the foundation up front and let the developer deploy into it. Then the harder question. The leaders making these calls today are the same engineers who lived through every prior cycle of this exact pain. Why are they letting another generation eat it again? Viktor's answer is one line: "It's my time now, baby." Ben does not disagree. PE pressure, VC timelines, race-to-market everything -- the budget exists, the tools exist, the patterns exist. What is missing is the will to invest two weeks up front so the last two months do not turn into panic. Ben's practical advice for any leader dipping a toe in: do not do it alone, inventory everything, talk to sales and finance and the developers, and assume the conversation you are having today will be obsolete in six months. Ben's contact information: LinkedIn: YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/41150165

DOP 348: Now It's Time to Panic 04/29/2026

DOP 348: Now It's Time to Panic Something flipped this year. Chatbots were a toy. Useful sometimes, but a toy. Agents are not. Agents take actions, hold credentials, write code, move Kanban cards, and run on cron schedules. The window between "this is interesting" and "this is existential" has closed faster than cloud, faster than Kubernetes, faster than any prior shift. Viktor's read is blunt. One person can now build a bigger business than most mid-size companies have ever managed. That is not hyperbole -- that is a description of what is already happening with a handful of solo-built projects shipping in weeks what used to take a hundred-person org years. The thesis: panic. Not because the sky is falling, but because larger companies cannot turn around overnight, and the gap between the people who get this and the people who are still scheduling meetings about scheduling meetings is widening every week. The conversation walks through what each big provider is actually doing. AWS is not pretending to compete on models -- they want the inference revenue. Microsoft is lost in Copilot button-stuffing. Google is quietly winning on three layers at once: TPUs, models, and inference infrastructure. Anthropic is on the path to becoming the next defining IPO, while OpenAI looks like a place to take money out of, not put more in. The Linux Foundation's new Agentic AI foundation got Anthropic's MCP, Block's Goose, and OpenAI's AGENTS.md spec. Viktor's reaction: those are heavy hitters donating not very much. Then it gets practical. Vendor-provided agents are like hiring a genius engineer who knows nothing about your company. Public skills are mostly nonsense -- if it is in public training data, the model already knows it; what is missing is everything specific to you, which is exactly what no public skill can provide. OWASP just published an Agentic AI Top 10 and most of it is least-privilege rebranded for agents. The cost story is also not what the marketing says: a 00 monthly subscription will not last a day for anyone working full-time with agents. There is a true story in here about a leaked token that turned a 00 monthly spend into 5,000 in two days. The hardest part of the episode is the part nobody likes hearing. If your output stays the same in 2026, you are in trouble. If you multiply your output, you are fine. Companies have always wanted to do more than they could afford to do. Now they can. The middle is where careers used to live. The middle is where the cuts are going. YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/41033815

DOP 347: Cozystack Turns Bare Metal Into a Managed Services Platform 04/22/2026

DOP 347: Cozystack Turns Bare Metal Into a Managed Services Platform #347: Andrei Kvapil has been around Kubernetes since the early days. Contributor to Cilium, Kubevirt, and a handful of other projects you probably use without realizing it. He is also the maintainer of Cozystack, a CNCF sandbox project, and the CEO of Aenix, the company behind it. The thesis: Kubernetes should be boring. Not exciting, not cutting-edge, not the thing everyone argues about. Boring like the Linux kernel is boring. Something that sits underneath everything and nobody needs to think about. Viktor takes it one step further and says it should be invisible -- developers should never need to know Kubernetes exists, any more than they need to know what kernel their laptop is running. Cozystack is Andrei's answer to a specific problem. ISPs, banks, finops shops, anyone in Europe who cannot or will not put their data in AWS -- they all want to offer managed databases, managed Kubernetes, object storage, the whole stack. Building that from scratch is hard. Running OpenStack requires a dedicated team that does nothing but tune networking. Cozystack bundles the pieces (Kubevirt, CloudNative Postgres, Cilium, etc) into one product with an aggregation API layer on top of Kubernetes itself. Helm becomes the extension language. The platform becomes a product. Then the conversation takes a turn. Andrei is the CEO of a bootstrapped company and he says flatly that without AI the company would not exist. Claude Code is moving Kanban cards. Clients send files generated by their AI agent and Aenix feeds those files to their AI agent to generate the response. Andrei's only wish is for this middle step -- him -- to stop existing. Let the agents talk to each other and call him when something actually matters. There is a hiring question in here too. If the next generation of engineers starts their career with AI on the first commit, do they ever build the mental model that lets them guide the agent when it goes wrong? Andrei thinks you still need deep understanding for anything serious. Viktor agrees. Speed versus quality is still a choice, and juniors who skip the "write it three times until it stops being garbage" phase are going to feel that gap eventually. Andrei's contact information: LinkedIn: YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/40959945

DOP 346: Fighting AI in Your Project Is a Terrible Mistake 04/15/2026

DOP 346: Fighting AI in Your Project Is a Terrible Mistake #346: Drive-by PRs, AI slop, maintainers burning out -- the open source world is having a meltdown and everyone wants to blame the robots. Viktor isn't buying it. The real problem started long before AI. Contributing to most open source projects has always depended on tribal knowledge and obscure docs nobody reads. AI didn't break that. It exposed it. When contributions were trickling in, you could get away with onboarding people via vibes. Now that contributions are a firehose, you can't. Viktor's take cuts in a direction that will annoy a lot of maintainers: your primary job is empowering contributors, not gatekeeping. And if a 20,000-line PR is drowning you, the answer isn't to block everybody. The answer is to change the whole review cycle -- because yesterday you were complaining about not enough contributions and today you're complaining about too many. That's a great problem to have. Solve it. Here's the part that will upset people. Viktor reframes what a developer's job actually is. If you think your role is typing on a keyboard, you're going to be disappointed. Your role is becoming a product manager. Asking the agent did you look at this, are you sure, what about that. Your job is no more. You just didn't receive the memo. There's also a thread running through the episode about auditing. Can you actually assess the health of an open source dependency you depend on? Viktor dares Darin to audit Kubernetes. Or curl. Or anything. Humans can't do it at all. AI can -- imperfectly, but better than nothing. Which means the old enterprise model (pay Red Hat, they'll handle it) starts to wobble when the value of someone else handling it drops because the tools can handle it for you. And there's a prediction. Right now when you ask AI to build something, it picks libraries based on training data. But what happens when the agent actually goes shopping -- analyzing projects, reading docs, deciding which dependencies to pull in? That changes the open source landscape in a way nobody is ready for. The episode ends somewhere quieter. Contributors, human or AI, should be cherished and trained over time. The hostility toward AI contributions is coming from maintainers who forgot that investing in new contributors is the job. The tools will make a mess. Then they will make less of a mess. Eventually we will be arguing about whether the feature should exist at all -- not whether the code compiles. YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/40818295

DOP 345: From Chat Prompt to Working Software with Kiro 04/08/2026

DOP 345: From Chat Prompt to Working Software with Kiro #345: Vibe coding works fine until your project gets complicated. That's the gap Amit Patel and his team at AWS built Kiro to fill. The tool launched with about six people in mid-2024, hit GA around October 2025, and the team still fits in a single room -- maybe a seven-pizza team by Darin's math. The core idea is spec-driven development, but not the kind where business analysts disappear for five years and come back with a document nobody needs anymore. Amit's version: you tell the agent what you want in a chat prompt, it writes the spec for you, and you iterate on it. Twenty minutes of back and forth and you've got requirements, a design, and a task breakdown. Then the agent executes. Two to three days later, working software. Here's where it gets interesting. Amit frames the human role as bookends. At the front, you define intent -- what needs to exist and why. At the back, you verify that what got built actually matches. Everything in the middle? That's where the tooling lives. And that middle is getting wider every month as agents run longer, handle more turns, and start working in parallel. But the gap between 'I can build it' and 'I built it right' is real. Amit's S3 example nails it. Ask an LLM to build a file upload app and you'll get one that works. Encryption at rest, encryption in transit, KMS, bucket policies -- none of that shows up unless you know to ask for it. The LLM will generate all of it on request. It just won't volunteer it. That's the experience gap, and it's why junior developers still need to become senior developers the old-fashioned way. One story that landed: a product manager on Amit's team used Kiro to go from conversation to working prototype overnight. Not a wireframe. Not a doc. A demo the engineering team could put into production. The roles aren't disappearing -- they're getting more fluid. The value was never in the writing. It was always in knowing what needed to be built. Kiro is now widely adopted inside AWS, with both an IDE and a CLI. Where it's headed next: agents that run in the background, handle multiple tasks at once, and get verified with formal methods instead of just hoping the output is right. But Amit's honest about the limits -- steering file adherence is, in his words, an art in itself. Non-deterministic LLMs will ignore your rules sometimes. Just like humans. Amit's contact information: LinkedIn: YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/40715350

DOP 344: KubeCon EU 2026 Review 04/01/2026

DOP 344: KubeCon EU 2026 Review #344: Kubernetes is boring now. That's the whole point. KubeCon EU 2026 in Amsterdam -- likely the biggest KubeCon ever at more than 13,000 attendees -- made one thing extremely clear: the container orchestrator is done being interesting on its own. Every keynote, every new sandbox project, every vendor announcement pointed the same direction. AI. Inference. Agents. NVIDIA donated a DRA driver for GPUs to CNCF. Google open-sourced their cluster autoscaler and shipped a DRA driver for TPUs. Red Hat brought LLM-D for disaggregated inference. NVIDIA contributed the KAI Scheduler for AI workloads. The Gateway API now has an inference extension in beta -- model routing baked directly into the Kubernetes networking layer. And here's the thing Whitney pointed out that should make everyone pause: you can't even run inference workloads in containers. They can escape. You need micro VMs. So the container orchestrator is orchestrating things that aren't containers. The platform engineering conversation shifted too. The bottleneck isn't technology anymore -- it's culture. Getting teams to work together differently. And if your company can't trust its own employees to make decisions, good luck trusting agents. Viktor's take on the determinism objection was blunt: agents aren't deterministic, but neither are you. You just think you are. One thread that kept surfacing: agents as first-class platform users. Not agents doing agent things -- agents as the users your platform serves. Viktor sees it in real time -- pull requests created by agents, reviewed by his Claude, responses written by the submitter's agent. Humans aren't even in the conversation anymore. The new CNCF sandbox projects tell the story too. LLM-D, KAI Scheduler, Higress (AI-native gateway). And then Velero -- the Kubernetes backup tool that everyone assumed was already CNCF -- finally donated by Broadcom. Which raises a fair question: is CNCF becoming a dumping ground for projects companies don't want to maintain? Probably some of both. Viktor compared the current state to the first five years of Kubernetes -- everyone focused on low-level components, trying to figure out how to combine 57 different tools. The next wave will be higher-level platforms that bundle all of it. And somewhere underneath it all, the mainframe keeps running. Viktor's bet: it'll outlive AI. YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/40653325

DOP 343: Your APIs Were Never Built to Be the Front Door 03/25/2026

DOP 343: Your APIs Were Never Built to Be the Front Door #343: Here's the thing about your company's APIs -- they were built for your own engineers to use inside your own software. Nobody designed them to be the front door. But that's exactly what's happening. Matt DeBergalis, CEO of Apollo GraphQL, makes a pretty compelling case that AI agents are turning internal APIs into the actual interface between companies and customers. Not the website. The APIs themselves. And most of them aren't ready for that. At all. Think about what happens when you point a model at a typical REST API. GitHub's API returns hundreds of fields for a single repository object. Fine when another service is calling it. But a model? All those extra fields are context you're paying for, and they make the model hallucinate. Matt says you need something between the model and all those backend services -- an orchestration layer that takes one request and handles the mess underneath. That's where GraphQL comes in. He draws a parallel that'll land immediately if you've been in this space a while. APIs right now are pets -- handwritten, named, carefully managed. But AI-generated code is about to produce way more microservices, which means way more APIs. They're going to become cattle. And just like containers needed Kubernetes, APIs are going to need declarative infrastructure to manage them at scale. The conversation takes an interesting turn when Darin pushes back on the idea that developers are becoming architects. His take: we're becoming product managers. Matt says both. Viktor throws in code reviewers. Matt's own story backs it up -- he codes more as CEO than he did as CTO, because AI handles the parts he never had time to learn. He doesn't know modern React. Doesn't need to. One more thing that should make any tech company uncomfortable: if AI agents are how customers find you now, what happens to your docs-page-driven acquisition funnel? Apollo's already made the shift -- their first audience for documentation is the models, not the humans. Matt's contact information: LinkedIn: YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/40515580

DOP 342: Your Company Documentation Is Useless for AI 03/18/2026

DOP 342: Your Company Documentation Is Useless for AI #342: Most companies have plenty of documentation. The problem is almost none of it is findable, current, or true. Between what's documented, what's actually true, and what people actually do, there are gaps wide enough to kill any AI initiative before it starts. Viktor makes a distinction that reframes the whole problem: there are two types of documentation. Why something was done -- that's eternal. How something works -- that's outdated the moment someone changes a config and forgets to update the wiki. The information about that change probably exists somewhere -- in a Zoom recording, a Slack thread, somebody's head -- but it's not where anyone would think to look for it. The running system itself is the most accurate documentation any company has. Your Kubernetes cluster tells you how many pods are running right now. Git tells you how many you wished you had. Those aren't the same thing, and pretending Git is the source of truth is a comfortable lie most teams tell themselves daily. RAG won't save this. Not the way most people imagine it -- point an agent at your docs and let it answer questions. That fails for the same reason Google's old enterprise search appliance failed. What could work is a continuous process that watches every information source, extracts what matters, and updates a central location intelligently. We have the pieces for this. Nobody's built it yet. The practical path forward: audit what you have before building anything new. Instrument your documentation the way you instrument applications -- find out what people search for and can't find. Design for retrieval, not storage. Build feedback loops. And stop treating documentation as a project with an end date. The companies that treat this as a strategic advantage instead of a chore are the ones that will actually make AI work for them. YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/40360050

DOP 341: AI Widened the Highway but Nobody Rebuilt the Bridge 03/11/2026

DOP 341: AI Widened the Highway but Nobody Rebuilt the Bridge #341: Nobody's arguing about whether you need feature flags in 2026. That debate ended years ago. But the code flowing through those flags? That's a different story. AI is writing more of it than ever, review times are climbing, and delivery throughput has actually declined. Trevor Stuart, co-founder of Split.io and now running Feature Management & Experimentation at Harness, calls it the six-lane highway ending in a two-lane bridge. The bottleneck didn't disappear. It moved. Coding got faster, but everything downstream -- reviews, security scans, delivery pipelines -- stayed the same width. Viktor points out this is the exact same pattern from the early agile days: his team shipped every two weeks, but testing still took six months. Different era, same structural problem. Feature flags are part of the fix, but not the way most people use them. Teams are now stuffing prompts, token limits, and temperature settings inside feature flag configurations and running A/B tests on AI agents in production. That's a long way from changing button colors on a marketing page, which is where experimentation started 15 years ago. The culture problem is harder than the tooling problem. Trevor has watched teams run one experiment, see it fail, and quit experimenting entirely. The fear of admitting failure kills more experimentation programs than bad data ever will. Meanwhile, the companies getting real results -- a fast food chain generating millions from kiosk experiments, a global bank driving hundreds of millions in customer acquisition -- are the ones treating experimentation as a permanent operating model, not a one-off project. The conversation also covers Trevor's path from co-founding Split to running it inside Harness post-acquisition. He stayed -- which doesn't happen as often as you'd think. Harness runs what he calls a 'startup within a startup' model, and he breaks down what that actually looks like from the inside, what was hardest to let go of, and why finding your 'why' matters more than any exit. Trevor's contact information: LinkedIn: YouTube channel: Review the podcast on Apple Podcasts: Slack: Connect with us at: /episode/index/show/devopsparadox/id/40287620