The questions worth sitting with
How Not to Be Very Powerful Broken People
A tool amplifies whoever holds it. As we pick up the most powerful tools in history, the decisive variable is not the machine but the human. Why becoming fully human is the work of this age.
Operational Productivity, Part 4: How Companies and Tools Must Rethink
The operational productivity AI playbook: move your budget, tools, and attention from the boardroom to the point of work, and build agents for the doer.
Operational Productivity, Part 3: The Ideas Only Possible With AI and Agents
The out-of-box ideas only possible with AI agents in operations. Put intelligence at the point of work so the doer can change the outcome in real time.
Operational Productivity, Part 2: The Dashboard Arms Race
Forty years of software made reporting brilliant and the work itself barely moved. Inside the dashboard arms race and the productivity paradox behind it.
Operational Productivity, Part 1: We Automated the Report, Not the Work
For forty years software made managers better at seeing the work. Operational productivity, the work itself, barely moved. Here is why that gap happened.
Peptides: The Molecules I Can't Stop Thinking About
A curious technologist's take on peptides — from GLP-1's metabolic revolution to experimental recovery compounds and AI-designed drugs.
The Second Brain That Works While You Sleep
How I built a local agent swarm that processes raw notes overnight — scouts, catalogers, critics, editors — and leaves a morning briefing in plain Markdown.
Being Human in 2035
Elon University's Imagining the Digital Future Center canvassed nearly 300 global experts on what AI changes about being human. The answers were less reassuring than you'd hope.
The Seventeen Percent Problem
McKinsey's 2025 workplace AI report found a 3x gap between what leaders think employees are doing with AI and what employees are actually doing. That gap is the real crisis.
Seven of the Top Ten Skills Are Human
The WEF Future of Jobs Report 2025 said the quiet part loud: the skills that matter most in the AI era are the ones machines can't replicate.
Learn Anything Faster: The Science of Rapid Skill Acquisition
A practical playbook for rapid skill acquisition: deconstruct any skill, drill the high-leverage 20 percent, and reach useful proficiency in weeks.
Graphs and State Machines for Agents, Part 4: Agents as State Machines
AI agent reliability is not a smarter prompt. Model the agent as a finite state machine and the safety-critical routes get enforced by a table, not a sentence.
Graphs and State Machines for Agents, Part 3: Building the Machine
A state machine implementation is a lookup table plus a tiny interpreter. Treat the transition function as data and whole classes of bugs stop existing.
Graphs and State Machines for Agents, Part 2: Every Graph Is a State Machine
The agent world keeps renaming one idea: chains, loops, graphs. Under every name sits a finite state machine, the 1950s formalism that already runs your CPU.
Kimi K3 and the Moment China Took the Open-Weight Lead
The Kimi K3 open-weight model from China is the largest ever at 2.8T parameters, single-digit points off the closed frontier at a fraction of the price.
Graphs and State Machines for Agents, Part 1: From Lines to Graphs
Most multi-step agents are a straight line that waits on itself. Model the work as an AI agent graph of nodes and edges and it collapses into a wider shape.
Thinking Machines Bet You Want to Own Your AI, Not Rent It
Thinking Machines released Inkling, an open-weight AI model built to fine-tune and own, not rent. Why the build-vs-rent call just changed for enterprises.
The Software Factory: When the Lights Go Off
AI coding agents turned engineering into running a software factory. The one thing that decides whether it helps or wrecks you is verification, not generation.
Reading the Mind We Made, Part 3: When the Microscope Becomes a Steering Wheel
AI interpretability's real payoff is control: seeing inside a model lets you edit inside it. Why legible internals are the only form of AI alignment that scales.
Reading the Mind We Made, Part 2: Two Ways to Read a Mind
Two roads lead into the machine. Frontier labs read their own closed models from inside; open source AI interpretability makes the microscope public.
Reading the Mind We Made, Part 1: The Science of Seeing Inside AI
We do not program modern AI, we grow it. AI interpretability is the new natural science built to read the billions of minds nobody wrote. Here is how.
Operational Productivity, Part 4: How Companies and Tools Must Rethink
The operational productivity AI playbook: move your budget, tools, and attention from the boardroom to the point of work, and build agents for the doer.
Operational Productivity, Part 3: The Ideas Only Possible With AI and Agents
The out-of-box ideas only possible with AI agents in operations. Put intelligence at the point of work so the doer can change the outcome in real time.
Graphs and State Machines for Agents, Part 4: Agents as State Machines
AI agent reliability is not a smarter prompt. Model the agent as a finite state machine and the safety-critical routes get enforced by a table, not a sentence.
Operational Productivity, Part 2: The Dashboard Arms Race
Forty years of software made reporting brilliant and the work itself barely moved. Inside the dashboard arms race and the productivity paradox behind it.
Operational Productivity, Part 1: We Automated the Report, Not the Work
For forty years software made managers better at seeing the work. Operational productivity, the work itself, barely moved. Here is why that gap happened.
Graphs and State Machines for Agents, Part 3: Building the Machine
A state machine implementation is a lookup table plus a tiny interpreter. Treat the transition function as data and whole classes of bugs stop existing.
Graphs and State Machines for Agents, Part 2: Every Graph Is a State Machine
The agent world keeps renaming one idea: chains, loops, graphs. Under every name sits a finite state machine, the 1950s formalism that already runs your CPU.
AI Is Wildly Overleveraged. It May Also Be Necessary.
The AI investment bubble is real and so is the demand behind it. A clear-eyed look at the capex, the debt, the circular financing, and whether revenue arrives in time.
Kimi K3 and the Moment China Took the Open-Weight Lead
The Kimi K3 open-weight model from China is the largest ever at 2.8T parameters, single-digit points off the closed frontier at a fraction of the price.
Graphs and State Machines for Agents, Part 1: From Lines to Graphs
Most multi-step agents are a straight line that waits on itself. Model the work as an AI agent graph of nodes and edges and it collapses into a wider shape.