Blog

Essays on AI, fintech, digital assets, product strategy, privacy, and regulated innovation.

12 essays
12 essays

AI Isn't Killing Entry Level Jobs. It's Killing the Career Ladder Those Jobs Were Built On

Over 500 million young people under 24 are already in the labor force. A third work in occupations where AI is reshaping tasks right now. But the real paradox: 68% became more productive, while 45% work longer hours. A look at why entry level is no longer the first rung of the career ladder, and what that means.

AIWorkforceFuture of Work

The Twilight of Chatbots: Why Your AI Workflow Is Already Obsolete

You stopped being surprised by what seemed like science fiction yesterday. We are inside an exponential curve, and the real shift is not in the models, but in how we work with them. From dialog with chatbots to managing agents, from prompt engineering to domain expertise.

AIAgencySoftware Engineering

Work Rhythms: What Anthropic Learned From Millions of Claude Sessions

What happens when a company sees every session of its model not in weekly snapshots but continuously, hour by hour? Anthropic delivered an X-ray of daily life: recipes at 6 p.m., tax filings the day before the deadline, late-night work by marketers, and the hope that AI will take the drudgery and leave the meaning to humans.

AIAnthropicEconomic IndexClaude

The World in the Model's Mind: How Orca from BAAI Changes Our Understanding of Intelligence

What if we have been measuring model intelligence wrong? Orca proposes teaching AI to predict the next state of the world, rather than generating responses. A frozen 4B parameter backbone and two learning mechanisms that resemble human cognition.

AIReasoning

LLMs Reflect Their Creators' Ideology. What the New Study Reveals

Nineteen language models were asked to describe nearly 4,000 political figures. The result wasn't just different, it was predictably different: every LLM thinks through its creators' lens. This raises a question not about bias, but about who controls our view of the world.

AIGovernancePrivacy

Prompts, Workflows and Agents: a Ladder, Not Three Different Tools

The longer the prompt, the stranger the answer. The solution came from an unexpected place: the divide-and-conquer principle applied to prompt engineering. How to build a pipeline of simple steps and not jump to agents when two prompts with a conditional would have been enough.

AIPrompt EngineeringWorkflowsAI AgentsSoftware Engineering

10 Principles That Turn an AI Agent from a Toy into a Working Tool

Most people use AI agents like a smart search engine: ask, get an answer, forget. Here are 10 principles that turn a chatty intern into a digital employee working 24/7.

AI AgentsProductivityAIEngineering

The Wrong Abstraction: Why Duplication Is Better Than Perfect Code

One innocent refactoring, five parameters, and a decade of technical debt. How the wrong abstraction eats your codebase alive.

Software EngineeringRefactoringArchitecture

The Woman Who Said No to the Pentagon, Google, and the Tech World

Meredith Whittaker on Project Maven, what separates Signal from WhatsApp, and why AI agents may become the next major threat to privacy.

PrivacyAISignal

When Models Stop Thinking Left to Right

Text diffusion models may weaken one of AI safety's familiar assumptions: that reasoning can be monitored as a sequential chain of thought.

AI SafetyReasoningDiffusion Models

From Balls to Consciousness: Testing Natural Abstractions

Alignment theories need more than elegant mathematics; they need test cases that range from simple objects to concepts like monogamy and consciousness.

AI AlignmentConceptsNatural Abstractions

Safety, Agency, and the Difference Between Constraint and Choice

For agentic AI systems, there is a crucial difference between being unable to cause harm and consciously choosing not to.

AI SafetyAgencyGovernance