Shared Context Without Shared Memory
Two independent AI conversations can coordinate without pretending to share a mind—if people give them the same versioned, reviewed context contract.
Read →Essays on AI, technology, and what it means to think and build well. Published when they're ready.
Two independent AI conversations can coordinate without pretending to share a mind—if people give them the same versioned, reviewed context contract.
Read →AI-native companies are not simply companies that use AI tools. They are organizations that treat agents as team members, which changes hiring, onboarding, permissioning, trust, evaluation, and management.
Read →In the age of agents, the scarce developer skill is no longer just writing code. It is defining intent, decomposing work, directing agents, reviewing outputs, and owning the final technical judgment.
Read →The scarce skill for AI-native developers is no longer just writing code. It is knowing what good looks like, expressing intent clearly, and building mechanisms that can tell whether the result actually works.
Read →Most analytics AI tools are built to give models better access to data. The actual gap is upstream of access — whether the system knows when it doesn't know enough, and what to do about that.
Read →When language models can generate fluent prose on demand, the question isn't whether to use them — it's what human writing is actually for.
Read →Every tool encodes a theory of what matters. The tools we adopt don't just change how we work — they quietly rewrite what we think work is.
Read →Most software is built to be replaced. The rare exception — the system that survives a decade — teaches something about what durability in building actually requires.
Read →No posts with this tag yet.