Context Clues
How to keep AI agents building what matters, not just what compiles.
- No. 47Why coding agents ignore your rulesYou wrote the rule down and the agent broke it anyway. It is not defiance, and a longer file makes it worse.10 min
- No. 46Context, comparedNine ways teams give coding agents context, scored against what independent research says a solution has to do.12 min
- No. 45Path dependenceThe choice your agent makes on the first day does not stay a choice.8 min
- No. 44Goodhart’s lawWhen the score becomes the goal, green tests stop telling you whether the product is right.12 min
- No. 43Gell-Mann amnesiaYou catch the agent’s mistakes where you are an expert and trust it everywhere else.9 min
- No. 42Tacit knowledgeSome of what your team knows cannot be written down, only shown.9 min
- No. 41Principal and agentEconomics spent fifty years on agents whose interests differed from yours.9 min
- No. 40The default answerWhen the same model hands you the same answer three times, it looks like consensus.9 min
- No. 39The whole productProduct-led growth gets you into one team. The whole product is what lets you spread through an enterprise.12 min
- No. 38Chesterton’s fenceDeleting has become nearly free. Understanding what you are deleting has not, and that gap is widening.8 min
- No. 37Survivorship biasWhy the evidence that decides your product is the evidence your dashboards structurally cannot show you.7 min
- No. 36Four risksYour coding agent has mastered one of the four risks that decide whether a product works.7 min
- No. 35Common knowledgeWhy a fact everyone already knows can change everything the moment somebody says it out loud.6 min
- No. 34Ghost decisionsWhy the code your agents generate is full of choices no one made, and what it costs to leave them unowned.11 min
- No. 33Productivity is not valueWhy record AI spend and reported productivity gains still do not add up to visible business value, and what the missing measurement is really about.8 min
- No. 32Your CLAUDE.md is a junk drawer: a migration playbookStandardizing the filename did not fix the file.11 min
- No. 31The firehose problem: your coding agent has all your permissionsWhy connecting your agent to raw GitHub, Slack, and Drive is the real security bug, and what to expose instead.10 min
- No. 30Loop engineering forgot the loop needs a memoryThe control flow was never the hard part. What the loop carries between iterations is.8 min
- No. 29You can’t review your way out of a context problemWhy scaling code review will not close the write-to-ship gap, and what will.10 min
- No. 28Everyone Is a Builder NowProduct management is dead.6 min
- No. 27Decision capture for AI coding agents: a write-path problem, not a discipline problemWhy documentation loses to the deadline at enterprise scale, and the write path that does not.9 min
- No. 26The context window is not the bottleneck. Relevance is.Why a ten-million-token window makes selecting the right decisions harder, not optional.7 min
- No. 25On TasteOr, tech companies try to speedrun philosophy.9 min
- No. 24Your agent stores beliefs, not decisionsWhy CLAUDE.md, raw integrations, and memory.md all fail the same test, and the seven properties a product context layer must have.7 min