About
About Frank Sellhausen
Applicability-first AI governance, principal-led. The discipline that examines a bank — applied to the AI now running inside one.
“There is a category of decisions that cannot be made solely by AI. The duty to the people affected cannot be discharged by a system. No system answers for itself. A person does.”
Every AI decision that affects a person carries a duty. The duty to know what was decided, why, on what evidence, by whom. That duty cannot be discharged by the system that made the decision. A person has to answer for it.
Federal agencies, military programs, regulated financial institutions, and the primes that serve them already carry this duty institutionally. The work of this firm is to keep that duty intact as AI gets woven deeper into the decisions they make.
Who the work is for
Institutions that already carry the duty — and want it to survive the next wave of AI.
Federal agencies operating under M-25-21 and the frameworks downstream of it. Military programs deploying AI into decisions that touch service members and mission outcomes. Regulated financial institutions building toward the FS AI RMF and the next generation of model risk supervision. Primes whose work touches fundamental rights or public consequence.
Anywhere AI decisions land on real people and a person has to answer for them.
Three regimes. One discipline.
Stand between a powerful capability and the people it affects, and keep it honest about reality.
Federal bank examiner. As a Risk Management Examiner in Charge at the FDIC, I examined institutions from small community banks to multi-billion-dollar asset holders against federal standards. Led model risk and third-party risk reviews. Conducted IT and BSA/AML examinations. Briefed findings to bank management, boards of directors, and regulatory leadership. What that work built: the reflex to look at a document with skepticism before belief, and the discipline to put findings on paper that survive examination.
Builder inside a Fortune 50. Five years progressing across analytics, data engineering, and AI product management. My team built the account-level data layer that served as the single source of truth across the enterprise — reconciling a landscape of disparate systems that had been producing different versions of the same facts. I authored the Data Engineering Playbook for my organization (provenance, lineage, controls, data contracts) and later owned the AI use-case approval process for the product management organization, defining the requirements, documentation, and acceptance criteria for AI features built on vendor-supplied models. What that work built: a builder’s read on what governance has to actually look like inside the build, not stapled on top.
Sixteen years in uniform. U.S. Army Drill Sergeant and Observer-Controller/Trainer at a Combat Training Center — the unit type that originated the After-Action Review. Observed units operating under pressure, assessed performance against standards, delivered structured, evidence-based feedback. Functionally, the same discipline as a model risk review: observe, document, defend the assessment, hand back something the next team can act on.
Credentials
Active and verifiable on demand.
- M.B.A. — Mays Business School, Texas A&M University
- B.S. Computer Science — Western Governors University
- B.B.A. Finance — University of Wisconsin-Milwaukee
- Professional Certificates: Full Stack Development & AI/ML Post-Graduate Program — The University of Texas at Austin
- Project Management Professional (PMP) — Project Management Institute
Where the practice is sharpening
In progress. The standards are moving; so is the practice.
- AIGP (Artificial Intelligence Governance Professional) — IAPP
- ISO/IEC 42001 Lead Auditor
- M.S. Computer Science, HCI specialization — Georgia Institute of Technology
- Stanford LEAD Executive Program — Graduate School of Business
Selected writing
How the same oversight failures repeat across banking supervision and AI deployment.
AI Amplifies the Same Problem — In Two Regimes
Generative AI doesn't introduce a new kind of organizational failure. It speeds up and surfaces the failure you already had.
Read essay →
What a Bank Examiner Sees When They Look at an AI Program
Most AI teams see their program from the inside. An examiner sees it from the outside — and the view is different.
Read essay →
Why I Started Thinking About AI Like a Bank Examiner
The discipline I learned examining banks is exactly what AI programs are missing. Not bureaucracy. The actual discipline: evidence, accountability, traceability.
Read essay →
Next step
Looking for an examiner's read on your AI deployment?
For federal agencies and enterprise governance functions standing up AI use-case approval, model risk processes, or third-party AI oversight.
Request a Briefing