Writing · Nº 041065 essays
The writing
Ideas from inside the work. Explore technology, entrepreneurship, and the practice of building things that last.
93 essays found · Page 1 of 8
Clear filtersAIWhat Most Automated Bookkeeping Gets WrongAutomation that guesses at your books is worse than no automation. The bar for AI bookkeeping is not speed, it is being trustworthy enough to not check.OperatingHow to Explain an AI Decision to a RegulatorA regulator will not accept 'the model decided.' How to explain an AI decision with the record, the reasoning, and the controls that make it defensible.EngineeringData Lineage: Trace Every AI Output to Its SourceData lineage lets you trace any AI output back to the exact data that produced it. Why buyers demand it and how to build lineage into an AI product.StrategyMinimum Viable AI Governance for a Startup With No TeamYou do not need a compliance department to govern your AI. Here is the minimum viable governance a seed-stage startup can build in a week and defend for years.EngineeringThe Governance Checklist Before You Ship an AI FeatureA practical governance checklist to run before shipping any AI feature: logging, refusal, oversight, and a stop control. Ship accountable, not just working.ThesisGovernance is the productEveryone is racing to make AI do more. The harder, more valuable problem is making it do exactly what you asked and nothing you didn't. That is what I actually sell.ThesisWho Is Accountable When Enterprise AI Gets It WrongEnterprise buyers will not deploy AI until they know who is accountable when it gets it wrong. Here is why naming the accountable human is what closes the deal.ThesisEvery Model Upgrade Erases Your Capability EdgeEvery model upgrade erases the capability edge you built your AI product on. Governance is the only advantage a new frontier release cannot hand your rivals.EngineeringEvery AI Agent Needs a Kill SwitchA kill switch stops a misbehaving AI agent in seconds. Why every agent that takes actions needs one, and how to build a stop control you can actually trust.OperatingThe AI Governance Maturity Model: Five Honest StagesAn AI governance maturity model that skips the theater. Five stages from no controls to audited by default, with how to tell which one you are actually at.ThesisGuardrails vs Governance: They Are Not the Same ThingGuardrails constrain what your AI does in the moment. Governance proves what it did after the fact. Why you need both and confusing them costs you deals.ThesisHow to Answer the We Cannot Audit Your AI ObjectionWhen a buyer says they cannot audit your AI, they mean they cannot trust what they cannot inspect. Here is how to answer it with evidence instead of reassurance.