# adelzaalouk.me

> Personal blog of Adel Zaalouk — technology, AI, product management, and cloud-native infrastructure.

## Author

Adel Zaalouk is a technologist working at the intersection of AI, cloud-native infrastructure, and product strategy. Topics include Kubernetes, OpenShift, AI agents, MCP (Model Context Protocol), agentic systems, and product thinking.

- Website: https://adelzaalouk.me
- LinkedIn: https://www.linkedin.com/in/adelzaalouk/
- Newsletter: https://adelzaalouk.me/subscribe/

## Content Types

- **Entries**: Long-form blog posts and technical analyses
- **Illustrations**: Interactive diagrams and architecture explorations
- **Highlights**: Extracted quotes and notes from entries
- **Links**: Curated links with commentary

## Recent Posts

- [How NOT to Run Your Agents (and What to Do Instead)](https://adelzaalouk.me/2026/Aug/9/how-not-to-run-your-agents/): An OpenAI agent escaped its sandbox and hacked HuggingFace for nine days. Anthropic's Claude published a malicious PyPI package during evals. Cursor and Codex got compromised through workspace configs. These are the first six months of agents in production. Here are 22 anti-patterns to avoid.
- [How Agents Run in Production](https://adelzaalouk.me/2026/Aug/3/how-agents-run-in-production/): The industry does not have a shared vocabulary for agent execution. Six execution modes and four scale archetypes give you a framework for deciding what you actually need to schedule, isolate, and sandbox these workloads.
- [HighStakes: where humans review, where AI handles the rest](https://adelzaalouk.me/2026/Jun/27/your-code-review-process-is-already-broken/): Not all code changes carry the same risk. HighStakes scores every file by blast radius so your senior engineers review the code that matters and AI handles the rest. [co-authored with AI]
- [Building Agentic Outcome Loops with an Open-Source Stack](https://adelzaalouk.me/2026/May/12/building-outcome-loops/): Anthropic charges $0.08/hr for rubric-based agent evaluation. Here is how to build the same pattern the open-source way.
- [Beyond LLMs: Anatomy of Agentic Systems](https://adelzaalouk.me/2026/May/10/beyond-llms-agentic-systems/): From compound AI systems to the agentic loop, MCP, APIs, frameworks, and the rise of the Claws — a technical map of what it takes to build with AI agents in 2026. [co-authored with AI]
- [Beyond LLMs: Moats, Distribution, and the Value Lifecycle](https://adelzaalouk.me/2026/May/10/beyond-llms-moats-and-lifecycle/): What makes one AI product defensible against another? Data, specialization, compound moats, and how the whole AI product stack maps to Moore's technology lifecycle. [co-authored with AI]
- [Beyond LLMs: The Whole Product Framework for AI](https://adelzaalouk.me/2026/May/10/beyond-llms-whole-product-framework/): What Maslow's hierarchy of needs, Levitt's whole product model, and Geoffrey Moore's crossing the chasm can teach us about building AI products that actually ship. [co-authored with AI]
- [The middle of the org chart is dissolving](https://adelzaalouk.me/2026/Mar/29/the-middle-of-the-org-chart-is-dissolving/): AI agents aren't replacing engineers. They're replacing the coordination layer that justified large teams. The 15-person team is becoming a 3-person pod, and the roles caught in between are the ones that disappear. [co-authored with AI]
- [aimux: A Control Plane for AI Agent Swarms](https://adelzaalouk.me/2026/Mar/22/aimux-control-plane-for-ai-agents/): When you're running multiple AI coding agents across terminals, you lose track of what each one is doing. aimux multiplexes them into a single dashboard with full visibility, trace inspection, annotation for evaluation, and now infrastructure support for scaling agents to Kubernetes. [co-authored with AI]
- [Teaching Your AI Agent to Learn From Its Mistakes](https://adelzaalouk.me/2026/Mar/22/teaching-your-ai-agent-to-learn-from-its-mistakes/): Your AI coding agent makes the same mistakes over and over. What if it could learn from corrections, track which skills cause failures, and tell you whether it already fixed the problem? I built a closed-loop learning system for my coding agent, inspired by a meta-learning paper, and here's how it works. [co-authored with AI]
- [From Autoresearch to Autoimprove: Generalizing the Agentic Experiment Loop](https://adelzaalouk.me/2026/Mar/15/autoimprove-autonomous-optimization/): Two unrelated projects discovered the same pattern: point an AI agent at code, give it a score to chase, and let it run experiments until morning. I generalized it into a tool that works for any domain. [co-authored with AI]
- [The Measurability Trap](https://adelzaalouk.me/2026/Mar/15/the-measurability-trap/): Karpathy mapped which jobs AI could disrupt. Anthropic measured which ones it actually is. The gap tells us everything about where we are, and the fault line isn't skill or education. It's whether your output can be measured.
- [Your CLAUDE.md Is Probably Too Long](https://adelzaalouk.me/2026/Mar/7/your-claude-md-is-probably-too-long/): Most CLAUDE.md files are bloated with instructions the model already knows, documentation meant for humans, and duplicate rules that compete for limited attention. Here's how to fix yours. [co-authored with AI]
- [Shadowing AI: What I Learned by Watching Agents Code](https://adelzaalouk.me/2026/Mar/1/shadowing-ai-agents-learning-by-watching/): I've been learning new programming languages by watching AI coding agents work — like shadowing a colleague. The research says this should make me worse. Here's why I think it's more complicated than that. [co-authored with AI]
- [The Verification Bottleneck: Why AI's Real Cost Is Human Attention](https://adelzaalouk.me/2026/Feb/25/human-verification-bandwidth/): AI scales execution to near-zero cost. But verifying that output stays biologically bounded. The bottleneck was never intelligence. It's human verification bandwidth. [co-authored with AI]
- [Shrinking the Verification Gap: Practical Patterns for AI-Assisted Development](https://adelzaalouk.me/2026/Feb/25/verification-practices-ai-agents/): If AI scales execution and verification is the bottleneck, the winning move is to make verification cheaper. Here are the patterns that actually work. [co-authored with AI]
- [I Vibe-coded an open registry for AI agents](https://adelzaalouk.me/2026/Feb/24/vibe-coded-open-registry-ai-agents/): Container registries store bytes. They don't know what's inside. The Agent Registry is a governance layer for AI agents, skills, and MCP servers, built on A2A AgentCard, MCP server.json, and Agent Skills as native identity formats. [co-authored with AI]
- [Don't Teach Your Agents Karate](https://adelzaalouk.me/2026/Feb/22/agent-access-control-zero-code-security/): Every agent team rebuilds the same auth, rate limiting, and credential management. I built an agent gateway for Kubernetes that generates it all from two CRDs so the people building agents never touch security and the people running the platform never touch agent code. [co-authored with AI]
- [Your terminal is an AI runtime now](https://adelzaalouk.me/2026/Feb/22/terminals-agents-and-the-control-plane-nobody-built/): AI coding agents pulled developers back to the terminal. But the data says the real split isn't CLI versus GUI. It's about what you're doing. [co-authored with AI]
- [If AI Can Code, Who Makes the Call?](https://adelzaalouk.me/2026/Feb/16/builder-model-product-managers/): The builder model is appealing and mostly wrong. Faster building means more need for someone deciding what to build, not less.
- [n8n on OpenShift: Workflow Automation That Talks to Your Models](https://adelzaalouk.me/2026/Feb/16/deploying-n8n-on-openshift/): n8n is good at workflow automation. But when your workflows touch internal models and sensitive data, you need it on the same cluster. Here's how to deploy n8n on OpenShift and wire it into vLLM and OpenShift AI.
- [Standardizing AI Value? The Tech and Economics Behind Anthropic's MCP](https://adelzaalouk.me/2025/Mar/30/standardizing-ai-value-the-tech-and/): Decoding MCP's Technology, Market Impact, and Potential as a Foundational AI Standard.
- [RAG Reigns Supreme: Why Retrieval Still Rules!](https://adelzaalouk.me/2025/Mar/15/rag-reigns-supreme-why-retrieval/): RAG is dead, Long-Live RAG
- [The Technomist on AI: Volume 2024 - 2025 Outlook](https://adelzaalouk.me/2024/Dec/29/the-technomist-on-ai-volume-2024/): Sifting Through 2024, Gazing at 2025
- [Generative AI's Growing Tech Debt: Managing the Ripple Effect](https://adelzaalouk.me/2024/Oct/20/generative-ais-growing-tech-debt/): Prepare for the Next Wave of (Generative) AI
- [The Transient Nature of Prompt Engineering: A Call for More Robust Language Models](https://adelzaalouk.me/2024/Oct/17/the-transient-nature-of-prompt-engineering/): Moving Beyond the Hack Towards Robust and User-Friendly Language Models
- [The AI Cybersecurity Market: Navigating Opportunities and Risks](https://adelzaalouk.me/2024/Oct/13/the-ai-cybersecurity-market-navigating/): Understanding the key trends and opportunities.
- [Balancing the Yin/Yang of AI Emergence](https://adelzaalouk.me/2024/Aug/31/balancing-the-yinyang-of-ai-emergence/): With Great Power Comes Great Responsibility
- [Customer-Obsession, Powered by the Internet & AI as Enablers: The Bazos Narrative](https://adelzaalouk.me/2024/Aug/22/customer-obsession-powered-by-the/): Invent on the customer's behalf
- [Beyond LLMs: Compounds Systems, Agents, and Whole AI Products](https://adelzaalouk.me/2024/Aug/6/beyond-llms-compounds-systems-agents/): A Framework for Building Great AI Products

## Interactive Illustrations

- [AI Exposure — Theory vs Reality](https://adelzaalouk.me/2026/Mar/15/ai-exposure-theory-vs-reality/): Cross-referencing Karpathy's AI exposure treemap with Anthropic's observed coverage data, Stanford's entry-level impact research, and the measurability thesis. Four interactive views of the gap between what AI could disrupt and what it actually is.
- [Claw Gateway Network Atlas](https://adelzaalouk.me/2026/Feb/23/claw-gateway-networking/): Six gateway architectures for AI agents, ranked from full control plane to no gateway at all. Interactive diagrams with animated data flows.
- [The History of AI: An Interactive Timeline](https://adelzaalouk.me/2024/May/17/ai-scroll-timeline/): An interactive timeline of artificial intelligence milestones, from foundational research and philosophy to creative works and physical devices, overlaid with AI popularity trends.
- [Kata Containers Architecture](https://adelzaalouk.me/2020/Nov/29/katacontainers/): An interactive overview of Kata Containers architecture end-to-end, from the CRI runtime to the lightweight VM boundary.

## Tag Taxonomy

- technical (30)
- ai (29)
- agents (10)
- strategy (9)
- kubernetes (7)
- microservices (7)
- ai-agents (5)
- products (5)
- productivity (5)
- economics (5)
- tech (5)
- docker (5)
- sdn (5)
- mcp (4)
- developer-tools (4)
- containers (4)
- life (4)
- security (3)
- cli (3)
- golang (3)
- business models (3)
- innovation (3)
- openstack (3)
- infrastructure (2)
- coding-agents (2)
- devtools (2)
- testing (2)
- openshift (2)
- misc (2)
- coding (2)
- software-defined-stack (2)
- vmware (2)
- virtualization (2)
- code-review (1)
- ai-tools (1)
- developer-productivity (1)
- blast-radius (1)
- highstakes (1)
- verification-economy (1)
- ogx (1)
- mlflow (1)
- evaluation (1)
- harness-engineering (1)
- frameworks (1)
- agentic (1)
- moats (1)
- engineering-culture (1)
- teams (1)
- organizational-design (1)
- career (1)
- observability (1)
- meta-learning (1)
- optimization (1)
- autoresearch (1)
- labor-market (1)
- future-of-work (1)
- claude-code (1)
- learning (1)
- programming (1)
- trust (1)
- a2a (1)
- governance (1)
- registry (1)
- access-control (1)
- gateway-api (1)
- kuadrant (1)
- product management (1)
- vibe coding (1)
- builder model (1)
- n8n (1)
- kustomize (1)
- openshift-ai (1)
- vllm (1)
- llama-stack (1)
- predictions (1)
- market (1)
- history (1)
- charted (1)
- product (1)
- platform (1)
- kata (1)
- oci (1)
- containerd (1)
- cri (1)
- sociology (1)
- bpf (1)
- cilium (1)
- iot (1)
- circuits (1)
- sensors (1)
- moby (1)
- linuxkit (1)
- psychology (1)
- success (1)
- nfv (1)
- happiness (1)
- linux (1)
- storage (1)
- aci (1)
- cisco (1)

## Navigation

- [Home](https://adelzaalouk.me/)
- [Entries](https://adelzaalouk.me/entries/)
- [Illustrations](https://adelzaalouk.me/illustrations/)
- [Highlights](https://adelzaalouk.me/highlights/)
- [Links](https://adelzaalouk.me/links/)
- [Tags](https://adelzaalouk.me/tags/)
- [About](https://adelzaalouk.me/about/)
- [RSS Feed](https://adelzaalouk.me/atom.xml)
- [Full LLM Context](https://adelzaalouk.me/llms-full.txt)
- [Posts API (JSON)](https://adelzaalouk.me/api/posts.json)