AI Risk Assessment
Structured review of a proposed system against the harms it could plausibly cause, ranked by likelihood and severity rather than novelty.
Armin Partners NYC
AI you can put in front of a regulator.
Manhattan-based advisory for regulated firms deploying AI — with the failure modes documented as carefully as the business case.
Start a ConversationAbout Us
Most AI programmes stall in the same place. The model works, the pilot is convincing, and then someone in risk or legal asks how it behaves on the edge cases — and nobody has written that down.
Armin Partners NYC was built in Manhattan to answer that question before it is asked. Lucas Whitmore founded the practice after years on the diligence side of financial services, where the discipline of documenting what could go wrong is simply the cost of operating. That background shapes how we engage: we treat a model like a counterparty, and we assess it the same way.
We work with banks, insurers, asset managers and the vendors selling into them. Engagements are deliberately short and scoped to a decision — deploy, don't deploy, or deploy with these controls — because open-ended governance work protects nobody.
Our Expertise
Six areas where we do our most useful work for New York firms.
Model Governance
Inventory, ownership, approval gates and review cadence — the operating machinery that keeps a model estate defensible as it grows.
Regulatory Mapping
Translating SR 11-7, NYDFS expectations and the EU AI Act into obligations that actually apply to your specific deployment.
Vendor & Model Diligence
Independent assessment of third-party AI: what the vendor tested, what they didn't, and where the liability sits when it fails.
Audit & Evidence Readiness
Assembling the documentation trail an examiner expects, before the examination rather than during it.
Policy & Training
Usable internal policy and role-specific training, written so the people bound by it can tell what is permitted.
Our Approach
- We assess the downside firstThe business case is already written by the time we arrive. Our job is the half nobody has drafted.
- We map to actual obligationsGeneric AI principles satisfy no regulator. We work from the rules that bind your charter and your jurisdiction.
- We produce evidence, not opinionsA governance engagement should leave you with artefacts an examiner accepts, not a memo describing good intentions.
- We stay reachableNinety days of follow-up included, because the difficult questions arrive at the first model review.
Contact Us
Tell us what you're weighing up. We'll reply within one business day.