Regulator-grade governance by design
At Proximitty, information security, model risk management, and regulatory compliance come before anything else.
SR 11-7 aligned by design
Each agent is scoped as a discrete model with documented inputs, outputs, limitations, and intended use. Every reasoning step runs inside a deterministic workflow, so your MRM team can review conceptual soundness, reproduce results, and monitor ongoing performance the same way they validate any other model in your inventory. Model cards, validation artifacts, and challenger testing materials are produced by the architecture and available on request.
Your customer’s data stays your customer’s data
Proximitty is currently in accordance of all SOC 2 principles, where customer data is segregated by tenant, encrypted in transit and at rest, and never used to train shared models. Access is governed by least privilege, logged in full, and reviewed on a defined cadence.
Humans decide, agents execute
Every agent operates inside explicit guardrails that define what it can do autonomously, what it must surface for review, and what it must escalate. Decision boundaries are configurable per workflow, per role, and per risk threshold — and every material decision defers to a human reviewer by default. Aligned with OCC guidance on AI governance and third-party risk management.
Always‑on, regulator‑ready documentation
Proximitty maintains regulator-facing documentation your team can incorporate directly into exam responses and model inventory entries. Our subject matter experts are available during examinations at no additional cost.
Everything you need to know about Proximitty
AI agents that ingest documents, spread financials, monitor covenants,
and service every borrower, built for C&I, CRE and SBA loans.
Resources
