11 PIECES · 05 SHELVES

What we learn building and operating our tools.

READING ORDER01—11

SHELF 01 · 03 PIECES

Persistent state

An agent system rents its model and owns everything else: the record of what it has done, the judgments on its past work, the rules and checks that encode its craft, and its standing with the people it serves. The owned parts are the ones that accumulate value, and the model is the part to plan on replacing.

06→07What compounds in an agent systemThe model in an agent system is rented and interchangeable. The parts that accumulate value over time are the records, the working rules, and the standing the system earns with the people it serves.ESSAYJUL 20265 MIN 07→08Systems appreciate, prompts depreciateA prompt improves one call on one model and loses its value at the next release. The examples, checks, and recorded failures a model runs through gain value with every release, because a smarter model immediately upgrades an effective system.ESSAYJUL 20266 MIN 10→11Ideas with bodiesAn idea held in a head or a note cannot be found by a stranger, cannot take a payment, and cannot learn from silence. Agents have made it cheap to form the body an idea needs to meet its market, meaning an address, an inbox, an account, and records that persist.ESSAYJUL 20266 MIN

SHELF 02 · 01 PIECES

Measurement

Most of what agents send into the world gets no response. When each stage of the path from sending to reply carries its own measurement, a non-response identifies the stage that failed, and each round of sends becomes an experiment that improves the next one.

08→09Instrumenting agent workWhen an agent's work gets no response, the useful question is which stage failed. Each stage of an agent-driven process can carry its own measurement, so that a non-response identifies the exact step that needs a fix.ESSAYJUL 20265 MIN

SHELF 03 · 01 PIECES

Governance

The humans behind an agent system cannot review thousands of runs directly. Corrections recorded as written rules, and enforced by automated checks, carry their judgment into runs they never see, and the human role narrows to approval and correction.

09→10Encoding judgment as rulesThe humans behind an agent system cannot review a thousand runs directly. Written rules carry their judgment into runs they never see, and each review makes the rules more complete.ESSAYJUL 20266 MIN

SHELF 04 · 05 PIECES

Foundations

Intelligence is now a resource anyone can buy, and a bought resource does nothing until it has a structure to run through. An agent is one such structure, an externalized piece of computation with a mind and a body, and agency is what the structures exist to expand: the power of a person or an organization to cause what they intend.

01→02Intelligence is waterIntelligence behaves like water. It presses on every surface it touches, passes through whatever opening exists, and takes the shape of its container, so the container sets the quality of what comes out.ESSAYJUL 20265 MIN 02→03Search that does not tireAn agent loop is a search algorithm: propose an action, carry it out, read the result, adjust. Retrying is the part of search that exhausts a person, and it is the part a machine gets at the price of a model call.ESSAYJUL 20266 MIN 03→04Shipping is mergingAgents made it nearly free to open parallel branches of work. A branch becomes progress only when its result is merged back into the one thing being built, and the merge is the step that did not get cheaper.ESSAYJUL 20266 MIN 04→05The agent bodyA product that can do anything gives you nothing to optimize and nothing to trust. An agent becomes a product when it has a body: a boundary separating inside from outside, and parts that persist after the model call returns.ESSAYJUL 20267 MIN 05→06Nothing grown in isolation survives transplantA seedling raised indoors dies when it is moved outside, because everything about it was built for the room and not for the ground. Products, founders, and AI agents fail the same way when they grow against a private model of a market instead of the market itself.ESSAYJUL 20266 MIN

SHELF 05 · 01 PIECES

Economics

A startup is a temporary structure built to search for a working business, and the search has always been priced in human time. When agents supply the labor, the search is priced in compute instead, and markets too small, too thin, or too short-lived to repay a person's attention become worth searching.

11→01The granular startupSteve Blank called a startup a temporary structure that searches for product-market fit. When agents supply the labor, the search gets cheap enough for markets that could never repay a person's time.ESSAYJUL 20265 MIN