Blog Series
The Architecture of Work
A four-part structural survey of work: a three-layer model of economic activity, then every one of its 136 activities assessed against AI reasoning, robotics, and traditional automation — by activity, by industry, and by the kind of solution each one demands.
4 parts 149 min read
Most claims about automation are made one job at a time, which is why they rarely transfer. This series starts from structure instead: a closed set of reasoning primitives, thirteen recurring domain patterns, and the 136 concrete activities that instantiate them across the economy. Only then does it ask what AI reasoning, robotics, and deterministic automation can actually do to each one on a three-to-five year horizon.
The four papers are four cuts through the same material — the model, the activity-level assessment, the same assessments arranged by industry, and the clusters that appear when you read across all of them at once. Read in order, or start with whichever cut matches the question you brought.
Part 1 of 4
The Architecture of Work: A Three-Layer Structural Model
Every productive activity in the economy resolves into three levels: a closed set of reasoning primitives, thirteen recurring domain patterns, and the open-ended sector activities that instantiate them. The structural model the rest of the series is built on.
Part 2 of 4
Opportunity Assessment: AI Reasoning, Robotics, and Traditional Automation
Each of the 136 Layer 3 activities assessed in isolation against three levers — AI reasoning, robotics, and traditional automation — on a three-to-five year horizon, unconstrained by present deployment but bounded by realistic trajectories.
Part 3 of 4
Sector Atlas: AI, Robotics, and Automation by Industry
The same activity-level assessments reorganized by sector, so a reader working inside one industry can find their own work without reading the theory first. Each entry closes with its cross-industry pattern affiliation.
Part 4 of 4
Solution Archetypes: Grouping the Opportunities
Read across all 136 assessments, the opportunities collapse into seven reasoning groups, six robotics groups, and five automation groups — clustered not by cognitive structure but by what has to be built, bought, governed, and staffed.