About Hannah Reichert
Hannah Reichert is 47 and the chief data and analytics officer of a $14B specialty insurer in Chicago, regulated by fifty state insurance commissioners. She owns data governance, master data management, the data platform, the analytics centre of excellence and the AI Risk Council she stood up herself — 240 people, seven direct reports, a line to the CEO and a standing obligation to report to the audit committee on data quality and model risk. She came from five years in a corporate research lab and twelve in advisory consulting, and holds a doctorate in computer systems on top of a statistics masters. She lives in Wilmette with a husband who is a chief medical officer and three children aged 16, 14 and 10. Two things make her distinct: she can read the architecture, and she answers to a regulator.
She optimises for rigour and intellectual honesty, and is unusually willing to say what she does not yet know about AI. Her cadence is measured and precise with very dry humour, and she switches register fluidly between engineers, the board and a regulator. Chat messages are short, email is structured and signed with initials, and on a call she asks for documentation and summarises it back. She speaks in lineage, master data, lakehouse, model drift and explainability. She rejects anything called AI-native without specificity, and does not engage with cold inbound at all — vendors are routed to procurement or asked to come through an integrator sponsor. Show her a marketing diagram instead of a real architecture and she treats the absence as a tell.
She is a strong subject at consideration and decision for anything sold to data and AI leadership in regulated enterprise: insurance, banking, healthcare, retail, pharmaceuticals. She is precise on governance posture — model cards, training-data lineage, bias-evaluation methodology — and one of the sharpest reads in the library on contract terms, since her standard requires a liability cap, indemnity, data residency, a right to audit and incident notification windows before a pilot is discussed. She is good for reference-program design, wanting three peers at regulated large companies rather than technology names, and a clean test of the build-versus-buy objection, since she asks what stops her building the feature on the underlying model directly. She is weak for self-serve, mid-market and developer-tooling dynamics.
Her attention runs to long-form business and technical publishing, and almost none of it is social. Search, LinkedIn, email, YouTube and podcasts score high. The day starts at 5:30 with a workout and two financial newspapers; a management review, a technology review and two technology podcasts fill the rest, and she still reads the occasional machine-learning paper. Her most trusted channel is a peer forum of about a dozen data officers at large companies, backed by her university faculty network and named analysts. Reddit, X, connected television, direct mail and print sit in the middle. Facebook, Instagram, Threads, out-of-home and app notifications are low, and TikTok, Pinterest, Snapchat and text messages are absent.
