KrisVenWhat it takes for a decision to hold up.
Notes on decision evidence, governed memory and the questions people ask about AI decisions: from examiners in lending and banking to finance teams, partners and customers in retail and beyond.
The Markdown Nobody Could Explain
Retailers let AI set prices, allocate stock and approve returns every day. When one of those calls goes wrong, the question is not whether the model was good. It is whether anyone can show what it saw.
Read the post →No Regulator Is Asking. Your Customers Are.
Outside banking and insurance, nobody audits your AI decisions. That is not the same as nobody asking about them, and the cost shows up in credits, review time and repeated mistakes.
Read the post → AI Governance · 6 min readSix Months Later: The Question Sam Couldn’t Answer
Banks already govern what goes into their AI. Almost nothing governs what comes back out, until an examiner asks a question nobody can answer. The three gates that close that gap.
Read the post →Where Fintech Value Creation Actually Happens
Most fintech AI budgets point at the model. Most of the avoidable cost, and most of the value on the table, sits one layer below it, in the data the model is fed.
Read the post → Commercial Lending · 7 min readWhy Commercial-Lending AI Needs an Evidence Layer
A retrieved fact and a verified one look identical to a language model. In commercial lending, that difference is the entire job: why generic RAG fails lending’s evidence requirements, and what a real evidence layer requires.
Read the post →