Notes
If you want new posts by email, subscribe on Substack.
AI & automation
-
Don't Sell Your Client a Colleague Who Chats. Close the Gap Between Quote and Payment First
A Telekom MMS survey of German companies ranked thirteen AI agent scenarios, and nearly every one it put on the short list is people acting as glue between two to four systems. What the survey found, what to keep in mind when reading it, and three things a small business should measure before talking about agents.
-
Don't Give AI Your CRM Password: Workflows as Tools for Agent Permissions
When a small business lets an AI assistant touch its CRM, email or accounting system, nobody needs to hand over an admin password. Wrap each permitted action in a small workflow, add human approval and a full log, and 'AI risk' becomes a permission map you can draw.
-
SMEs and AI in 2026: Ten Sources, and the Question of What They Measure
Census, Eurostat, the OECD, the EIB, Bitkom, Goldman Sachs and others disagree on how many small businesses use AI and who benefits most. Lining them up shows where they agree, where they split, and where outside advisers can and can't help.
-
The Triple Debt Model: A Framework for Thinking Through Your Automation
Borrowing the technical, cognitive, and intent debt distinction to think about how Agents can actually lighten the load, instead of just moving the checking work onto a person.
-
Two Studies, Two Opposite Answers
Two studies on whether AI narrows performance gaps reach opposite conclusions. They reconcile once you separate the tasks — but both have a methodological catch worth naming first.
-
When AI Agents Form Teams: Redefining Organizational Design from Systems Theory
Drawing from control theory, information theory, and cybernetics, this article proposes a foundational framework for Agent System Design. The goal: to establish a discipline that does not yet exist.
EU rules, in practice
Judgment & work
-
Can Judgment Itself Be Surpassed?
From the internet-era knowledge analogy to chess's full arc, thinking through whether judgment might be the first thing this AI shift actually puts to the test.
-
Universities Can't Block AI, So They Changed the Test
NYU, the University of Chicago, the University of Sydney, and Cal State Maritime Academy are each rethinking how they test students, and each choice reveals a different bet.
-
Before Asking Whether AI Is Conscious, We Need to Know What Consciousness Is
Hinton says AI is already conscious. Before agreeing or disagreeing, there's a more fundamental question to face: what do we actually mean by consciousness?
-
AI Transition Infrastructure: What I'm Building and How It Connects
These projects look scattered from the outside. They're actually responses to the same underlying question: as AI disrupts the conditions beneath work, education, and knowledge circulation, some of society's connective interfaces are starting to crack.
-
Why I'm Starting an AI-Era Apprenticeship Project
AI is removing the economic condition that made entry-level training viable. This isn't something the market will fix on its own. It's a transfer mechanism that needs to be deliberately rebuilt.
-
Don't Just Copy Ponytail: Building a Development Methodology That's Actually Yours
Every few weeks, a new AI development methodology goes viral. The question isn't whether it's good. It's whether you can tell — for yourself.
Show 9 more
-
AI Made Us More Productive. So Why Is Everyone More Busy?
AI has boosted productivity, but almost nobody feels less busy. Employees are driven by fear, solo workers by opportunity cost — two very different situations arriving at the same destination.
-
When AI Starts Decomposing Work, Careers Get Sliced Too
AI may not simply eliminate jobs, but first decompose them into smaller responsibility modules. As companies start asking candidates about building and managing agent teams, careers themselves may be restructured and repriced.
-
Why Agent Teams Need Editorial Boundaries
Why agent teams need clear editorial boundaries, ownership, and separation of concerns as they grow from single assistants into collaborative systems.
-
When WhatsApp Shows 47 Unread Messages
Why we moved from real-time chat overload to a dual-track system with Notion as the passive knowledge channel. The story of how 47 unread messages led to a philosophical shift in AI agent communication.
-
Notion Integration: Building Passive Knowledge Channels for AI Agent Teams
Why more chatbots isn't the answer. How we designed a dual-channel system with WhatsApp for real-time and Notion for passive knowledge management.
-
Why Your AI Remembers Everything but Understands Nothing
From storage to curation: How we redesigned AI's memory architecture to evolve from a digital hard drive to an intelligent partner.
-
From Solo to Squad: How AI Agents Are Transforming Our Workflow
In the era of The Great Reordering, how do we evolve from solo players to human-AI collaborative teams? Introducing OpenClaw, Sub-agents, and our AI Squad.
-
Prediction is Cheap, Strategy is Key
Prediction is cheap; strategy is key. In the face of AI-driven uncertainty, we shouldn't gamble on a single future, but instead build an 'antifragile' portfolio of choices.
-
Why 'Beyond the Image'?
In the era of The Great Reordering brought by AI, the answers of 'it has always been this way' are failing. How should we position ourselves?