using agents to build agents

You describe it, the agent writes it: then it runs somewhere you can defend

Two halves of one idea. First, a coding agent that turns your design into working agent code, which only works if you give it a proper specification and the right domain knowledge, or it confidently writes against the wrong API. Second, a managed runtime that starts from a locked-down sandbox and lets you add capability on top, so “where does our data go and what can this reach” has an answer before anything ships. Nine sections, eight simulations, every trap that catches people in practice.

Try it: which end should you start from?

Try this: four questions about your situation. The answer is not which approach is better. It is which problem you would rather own, and the two are genuinely different questions.

in real lifeBuilding a workshop or renting a serviced unit. One gives you total freedom and a lease on the roof repairs; the other gives you rules and someone else's plumber.

start byanswer in the first meeting to the security question and watch the bar move.

words hereruntime the managed service that runs your agent and handles the loadframework code you write and host yourselfisolation keeping one customer’s work away from another’s

Try it: drill the scenarios instead of reading them

Try this: the same scenarios as below, one at a time, with the answer hidden. The alternatives are real answers to other scenarios on this page, so none of them is obviously silly, which is the point.

in real lifeFlashcards versus a revision guide. Reading the guide feels like learning and mostly is not. You recognise the answer when you see it, which is a different skill from producing it.

start bypress next scenario without reading the table below first. Recognising an answer and recalling one are not the same thing.

words herescenario a situation described the way an exam or an interviewer would put itdistractor a wrong option that is a real answer to something else