From Islands to Archipelago: Building the Shared Space Between People and Agents
Part 2 of The AI Cohesion Paradox. Part 1 named the problem. This part is about the habits and processes that fix it.
Your Agent Doesn’t Need an Ontology. It Needs a Conceptual Model.
Your agent does not need an ontology; it needs a conceptual model, and the two are not the same thing. A conceptual model is a much larger idea than a graph, and most teams discover this only after spending months on the wrong problem.
Agentic Graph Engineering for Agentic Coding: The graph you can generate and the graph you can’t
Graphify maps what your code is. TypeGraph records what it is for.
Markdown, Supercharged: XML Tags and YAML for Agent-Ready Specs and ADRs
Keep your specs and ADRs in Markdown, but add two layers on top: YAML for the facts a tool or an agent must read exactly, and XML tags for the boundaries a model must respect. You get most of the structure people now reach for HTML to provide, at a fraction of the tokens, and the files stay diffable and human-writable.
Strongtalk: The Fastest Smalltalk You Never Used
How a typed Smalltalk from a tiny 1990s startup ended up inside Java, Chrome and Flutter
The Kernel Is the Reviewer: Why the Future of Agentic Coding Is Verifiable Lean 4, Not Markdown
Humans own the statements. Agents own the implementations and proofs. The kernel arbitrates.
No, Self Is Your Father: The Language Behind Every JavaScript Object
A 1986 research language with no classes gave JavaScript its objects and V8 its speed. Then we spent twenty years pretending it didn’t.
What did the system know, and when?
A worked example from a regulated domain: sanctions screening at a payments firm. Facts have two clocks and the screening rules are λProlog programs stored next to the facts, so every audit question becomes a query you can run again, years later, and get the same answer.
Your Agent Remembers Things That Never Happened
Why LLM agents need a logic layer, what it fixes, and a gentle way to learn it
Debugging agents needs a past tense : Memory you can audit
Why agent memory that overwrites is a liability, and what debugging looks like when it doesn’t.
Your codebase has two graphs. You only maintain one.
A deep dive into typed graphs as a working tool for developers: what to model, what to leave to the machine, and where specs belong now that they live in the repo.
The Book Was About Space. The Year Was About Why.
A year ago this Wednesday I published a book with a slightly reckless claim: almost everything an AI agent needs to remember can be expressed with four relations. NEAR. LEADS TO. CONTAINS. EXPRESSES PROPERTY.
A Mile to the Mile: Lewis Carroll and the World Models of AI Agents
Philosophers for AI
Your Agent Is Guessing About Your Domain
A nullable column can’t tell your coding agent whether the rule was optional or nobody ever decided
Near Is Not a Distance
Your vector index signed a four-clause contract on your behalf. Amos Tversky showed all four clauses are false about human similarity — and a graph with a NEAR relation is what you get when you refuse
The Missing Shade of Blue
David Hume, the tyranny of resemblance, and the two relations your vector database threw away
Your Agent Is Not a Thing
Alfred North Whitehead, process ontology, and the events your knowledge graph is pretending aren’t there
The Hidden Art of Knowing: Kant’s Schematism and the Conditions of Possible Agency
Part of the series: Philosophers for AI
Everyone’s Building Loops. Almost Nobody’s Building the One That Remembers.
“Stop prompting your agents. Design the loops that prompt them.” It’s the right instinct — and it’s exactly half a loop. Here’s the other half: the artifact the loop runs on.
You Are Not Managing an Animal. You Are Summoning a Ghost.
Karpathy’s three layers — Spec, Verifier, Environment — are grounded in the actual papers and mapped onto a pipeline you can run Monday morning. Plus the one idea that makes the third layer make sense: a world model.
Your Repo Remembers Code and Forgets Why — Three Artifacts That Fix It
ADR, OpenSpec, and the lattice (lat.md/): a change-machine, a decision log, and the substrate that holds the vocabulary, both of them quietly assume.
Specs Are Just the Names We Give to Properties
Why “facts vs. specs” is the wrong fight — and how to wire OpenSpec to fast-check, proptest, and Hypothesis so your living spec actually compiles to something falsifiable.
The Token Economy of Coding Agents Is Broken — Here’s the Split That Fixes It
Why the next leap in AI engineering is not a smarter model. It is a quieter discipline about which model gets which tokens.
The AI Cohesion Paradox: Why AI-Empowered Teams Cooperate Less — And What Leaders Can Do About It
Something quietly changed in the organizations that moved earliest and fastest on AI. Their people got faster, their output went up, their sprint velocity looked great — and then the seams started to show. Departments that had once collaborated spontaneously began operating like sovereign states. Engineers who used to grab a coffee and talk through a design decision now just... asked their agents. The common context that used to live in a team’s informal exchanges — the “have you talked to Dave about this?” moments — started evaporating.
Code Is the Side Effect: Why I Wrote “The Clarity Engineer”
A book about how engineers, teams, and organizations actually transform around AI — not another pile of prompt templates.
Obsidian, Supercharged: The AI Revolution in Personal Knowledge Management
How a markdown vault became the operating system for the thinking machine age
The Scaling Wall: Moving Beyond MD Files in Multi-Agent Systems
Right now, files are everywhere in the agent ecosystem. Markdown files are the default for everything — skills, custom instructions, agent descriptions, commands, memory. The entire vibe-coding movement runs on the assumption that a folder of .md files is all you need.
When Your Body Stops Speaking: Tea Meditation as a Science-Based Burnout Recovery Protocol
How L-theanine, theabrownins, and your vagus nerve do what no meditation app can
The Office Guru: How the Tea Ceremony Transforms an Ordinary Workday into a Mindfulness Practice
Step-by-step 5- and 15-minute protocols to help you focus, reduce stress, and not say something you'll regret to a colleague
The Neuroscience of Kung Fu Tea Meditation: Why Small Batches Train the Brain
Most writing on Gongfu Cha leans on aesthetics — the kettle, the clay, the lineage. The neuroscience is more interesting and, for some people, more useful. The brain is a difference-engine: it lights up for novelty, for mismatch, for this sip not being the previous sip. Gongfu brewing — many small infusions from the same leaves, each one slightly different — is almost custom-built to feed that engine. And, importantly for people who can't safely turn attention inward, the practice grounds you through outside-in sensing (taste, aroma, warmth in the cup) rather than through interoceptive body scanning, which can backfire in chronic pain and trauma.
Wittgenstein for AI Agents
Why AI Leaders need to read philosophers more
The Lie of the Genius: Why Claude Code Tells You You’re Brilliant, and What That’s Doing to Your Brain
We have become heavily addicted to Claude Code. I’ll say it plainly, because polite hedging would already be a small act of self-deception, and self-deception is exactly what this piece is about.
A World Model for Agentic Coding
In robotics, the original “world models” were models of physics — they explained the environment to an agent so the agent could move, grasp, and not break things. NVIDIA built one. A lot of people built one. The shape was always the same: capture the laws of the environment so the agent can reason about consequences before it acts.
Why AI Memory Fails — And What Human Cognition Teaches Us About Fixing It
Every AI agent today faces the same silent problem: it forgets, it hallucinates, and it can’t tell you why it believes what it believes. The root cause isn’t a lack of data or compute — it’s a lack of structure. Specifically, the kind of multi-layered, cross-linked memory structure that evolution spent millions of years perfecting in the human brain.
Books vs Devices and Memory Mechanics
I was recently talking about a study involving various tests, like PISA, where everyone is criticizing the younger generation. The results were predictable because there is a major ongoing discussion now, for example in Germany, about banning social networks, though that is actually about something slightly different.
Ways of Working with AI: Why Leaving the Flow Matters
We live in a culture obsessed with entering the flow state. There is no shortage of focus music, productivity rituals, and concentration hacks — most of which, by the way, don’t work the way their creators promise. The brain does operate on specific electrical frequencies: alpha waves around 8–12 Hz associated with relaxed alertness, theta waves around 4–8 Hz linked to creative insight, and the famous gamma bursts around 40 Hz that accompany moments of cognitive integration. The problem is that most streaming audio formats and their compression artifacts destroy exactly those frequencies. Binaural beats require uncompressed audio delivered through headphones; what you find on most platforms is a degraded approximation that produces no measurable effect. So don’t waste your focus on the focus tools. That is already a kind of irony worth sitting with.
The “Twitterization” of the Mind: Why Gen Z is Struggling with Cognitive Tests - Is it AI false ?
There is an active discussion online right now about the idea that Gen Z is “stupid”. This stems from studies and standardized tests indicating that this generation significantly underperforms its predecessors on cognitive tasks. Those who are more tech-savvy attribute the problem to AI, whereas others point to devices, claiming that mobile phones and screen time are essentially destroying the brains of young people.
Generative AI and the Paradox of Intensification
Hey, it’s a typical Wednesday, and today we’ll talk about quite interesting research that focuses on how generative AI transforms your work. And in theory, you should work faster, do more, or work less; but as human nature says, that’s not the case. And instead of acceleration, you have the intensification, actually. So you work more intensively, and you’re more exhausted as a result.
Why Simula — Not Smalltalk — Changed the World: The Power of Simulation and Coroutines
When we trace the genealogy of modern programming, we often celebrate Smalltalk as the progenitor of object-oriented programming. Yet this narrative obscures a deeper truth: Simula, created by Ole-Johan Dahl and Kristen Nygaard in 1962–1967, fundamentally transformed how we model complex systems, and its influence extends far beyond the syntax of classes and objects that Smalltalk popularized.
The Deep Craft of Sweetness: Why M-Shaped People Matter
Or: What Making Tea Without Sugar Teaches Us About Technical Excellence
How Language Reshapes Time: When Your Mother Tongue Rewrites Reality
We treat language as a neutral tool — a transparent medium for transmitting thoughts that exist independently in our minds. But what if the structure of your native language isn’t just describing your reality, but actively constructing it? What if the words you speak are quietly rewriting how you perceive something as fundamental as time itself?
Liu Bao: The Ancient Tea for Modern Technical Leaders
In the world of fermented teas, Liu Bao occupies a unique position—older than Shu Pu-erh by over a millennium, yet less understood and more variable in quality. For technology leaders and architects navigating complex systems and high-stress decision-making, Liu Bao offers something rare: deep relaxation without loss of focus, a neurochemical state particularly valuable for sustained creative work.
From Survival Mode to Curious Mind: Why Leaders Must Reclaim Their Questions
The most dangerous thing happening in your organization isn’t the competition, market disruption, or technological change. It’s that your leaders have stopped asking their own questions.
The Attention Deficit: Why Innovation Struggles in an Age of Information Overload
I've spent considerable time studying pedagogical approaches—how to explain complex topics, manage cognitive load, break down intricate systems into digestible pieces. The conventional wisdom suggests that complexity itself is the barrier: if only we could explain things more clearly, simplify the diagrams, polish the analogies, people would engage with difficult ideas.
The Future of Agentic Vibe Coding: Constrained Natural Language and Language-Oriented Programming for DSLs on Steroids
The pronouncement that “programming is dead” and natural language is the new programming paradigm has become a familiar refrain in AI circles. But before we accept this proclamation at face value, we should remember who’s making it: those who’ve bet everything on AI and profit from this particular narrative. The reality, as usual, is more nuanced and far more interesting.
The Secret Bubuster: When Ancient Tea Meets Bulletproof Biochemistry
There exists a nearly forbidden recipe, developed by lost civilizations (or perhaps by sleep-deprived engineers in Berlin). I call it the Bubuster — and it's basically what happens when Tibetan butter tea meets Silicon Valley biohacking, with a dash of Slavic stubbornness.
The ai story ....
You professionally dig with a shovel. The shovel is simple: sometimes you use small shovels, sometimes big ones. And then one wonderful day everyone goes crazy and starts writing that shovels are dead and everything urgently needs to be tractorized.
Renting, Not Owning — and Why It Is a Problem for AI
Philosophical Wednesday reflection
What the Hell Are Architects Doing? The Architect’s Role from Startup to Enterprise
What the Hell Are Architects Doing? It’s a provocative question but an important one. The role of an architect in software development spans startups to massive enterprises, and it’s often misunderstood. Architects are like philosophers in the tech world—working through mental models, bridging communication gaps, and making sense of complex systems. This article will explore architecture, the different types of architects, and their roles in diverse organizations, from startups to enterprises.
Clarity Traders: Beyond Vibe Coding
The software engineering community is experiencing a moment of collective vertigo. AI coding assistants, from Claude Code to various copilots, promise that anyone — even founders without deep technical knowledge — can now build software as easily as recording a video. The dream of the “citizen developer” feels tantalizingly close. Yet this narrative fundamentally misunderstands what software engineers actually do.
Smalltalk: The Language That Changed Everything and Why It Still Matters for AI Agents
In the pantheon of programming languages, few have had such a profound impact yet remain so misunderstood as Smalltalk. Created in the 1970s at Xerox PARC by Alan Kay, Dan Ingalls, and Adele Goldberg, Smalltalk didn’t just introduce object-oriented programming to the world — it created an entire vision of computing that we’re only now beginning to fully appreciate in the age of AI agents.
Lepiter: A Moldable Knowledge Management System for the Age of Programmable Documents
In the landscape of knowledge management tools, most applications force users into rigid, predefined structures. You write notes in one tool, code in another, and documentation in a third. Lepiter, developed by the feenk company as part of the Glamorous Toolkit (GT) ecosystem, breaks this paradigm by offering something fundamentally different: a moldable development environment where knowledge management becomes an interactive, programmable experience.
1
1
Problem First: How to Make Hardware and Models Tailored for AI Agents
Today, let’s delve into problem-oriented approaches in technology. We will focus on the differences between computer engineering and computer science, general-purpose CPUs versus tailored hardware, hybrid solutions like FPGAs, and how these ideas can revolutionize AI hardware and models. Let’s extract key insights from these topics and explore how a “problem-first” mindset can transform AI agents.
From Spasticity to Serenity: How Tea and Yerba Mate Aid Muscle Recovery
My journey into the world of tea began in 2005 with a gourd of Yerba Mate. To be honest, it wasn’t "love at first sip" regarding the flavor, but the physiological effect was undeniable. I noticed a paradoxical sensation: sharp mental focus coupled with a subtle relaxation of my overstrained muscles—without the lethargy often associated with sedatives.
Turkish tea in Berlin apartment
Here's the translation:
What We Make of What We've Been Given: Sartre's Guide to Engineering
A Philosophical Wednesday Reflection
The Right to Be Misunderstood: Building the Future Before Its Time
A Philosophical Wednesday Reflection
Controlled Euphoria for Creativity: How Aged Sheng Pu-erh Tea Becomes a Serotonin Production Engine
As engineers, we're constantly seeking optimal states for deep work—that elusive combination of laser focus and creative flow. While most reach for coffee's caffeine jolt, a more sophisticated approach lies in understanding how aged sheng pu-erh tea transforms into a gut-brain axis optimizer that naturally elevates serotonin production.
Finding Ground: Ancient Teas for an Anxious Generation
We are an anxious generation. If anxiety overwhelms you, a psychologist or psychiatrist can certainly help—and I'm not offering medical advice here. The Swiss Alps and the collapse of Muscovy might also bring relief. But there's something closer at hand that deserves attention.
Why Engineers and Architects Need Tea Meditation — and How It Opens the Door to Flow
In the worlds of engineering and architecture, we often talk about systems, precision, patterns, constraints, and creative breakthroughs. But rarely do we talk about the inner conditions that make real insight possible. We obsess over tooling and methodologies, yet overlook a far simpler instrument of clarity: a cup of tea.
We do not solve the problems we have; we solve the problems we understand.
By Lene Rachel Andersen
Breaking New Ground: Journey into AI, Privacy, Sovereign and Edge AI
After two years of intensive research, countless YouTube videos, and numerous articles, a significant milestone has been reached in the intersection of artificial intelligence and privacy-focused computing. What began as philosophical discussions about engineering and AI has evolved into a comprehensive book series that tackles some of the most pressing questions in modern technology.
Agentic AI: What is Agent Identity About?
from identifiers to capabilities
AI-Native Organizations: The Future of Business
How AI native differ and why it is game changer
LOKA Protocol for AI Agents On Top of SSI
AI Agent protocol
Why We Build Our Memory System for AI Agents
When researching memory systems for AI agents, you’ll find numerous options on the market. Some are relatively simple, built around semantic graphs, while others are more document-oriented. Notable solutions like ZAP with its temporal graphs and Memento, Mem0 offer impressive capabilities. However, after careful evaluation, we determined that building our own memory system was necessary for several compelling reasons.
Forgetting in AI Agent Memory Systems
In the realm of artificial intelligence, particularly with conversational agents, we often focus on how to make systems remember more effectively. However, one of the most critical yet overlooked aspects is how these systems should forget information. As human memory doesn’t simply store and retrieve data, but dynamically prioritizes and sometimes forgets information, AI memory systems need similar capabilities to function effectively and naturally in human interactions.
Privacy-First AI Agents for Ethical AI
In today’s age of rapid technological advancements, privacy has become one of the most critical yet challenging topics in software design and artificial intelligence (AI). The concept of privacy often appears daunting due to the pervasive data collection practices of modern systems. However, with proper adherence to Privacy by Design principles, the task becomes more manageable. This article explores the need for privacy-first AI agents, the fundamental principles underpinning privacy, and the ethical implications for AI systems.
AI Agents Need Agent-to-Agent Protocols to Excel
The development of AI agents has been a booming topic in recent years. From their early conceptual stages to the sophisticated models we see today, the focus has been mainly on building robust agents. However, the importance of Agent-to-Agent (A2A) protocols has been largely overlooked. These protocols are crucial for enabling seamless interaction, fostering interoperability, and unlocking the full potential of AI agents. Here, we explore why A2A protocols matter and how they shape the future of AI.
Experts vs. Generalists vs. Barefoot Gurus in Startups: Does Fullstack Exist?
Happy New Year! Today’s philosophical discussion explores the enigmatic concept of the fullstack developer and its relevance in modern engineering and startups. We’ll delve into whether fullstack developers exist, the ongoing debate between generalists and experts, and emerging trends like hybrid ML/LLM fullstack roles. Using the structure of the subtitles provided, this article will also highlight engineering thinking and the evolution of barefoot experts.