RunlineRunline
Runline Overview

Governance-First

Compliance embedded in the architecture, not bolted on. Built for NCUA examination from day one.

Compliance is not a feature we add. It is the foundation we build on. Runline is built for NCUA examination from day one, and human oversight is enforced by the architecture, not by policy: the system cannot file a SAR or post a journal entry without a human approval on the record.

Built for the examiners
NCUAPart 748
BSA / AMLSAR-ready
GLBASafeguards
SOC 2audit-ready
Gated autonomy: a Runner's action passes through Allow, Ask, or Block, then executes, with every action written to an append-only audit log.

Gated autonomy

A Runner keeps full capability, with mechanical checkpoints at the exact points where a mistake would be expensive or hard to reverse. Three tiers, never two:

  • Allow: routine, auto-approved.
  • Ask: needs a Pilot's approval, and can still execute once approved.
  • Block: a hard stop, no approval path, surviving even elevated autonomy.

What NCUA requires, and how we answer it

NCUA requirementRunline's architecture response
Human oversight of AI decisionsApproval gates on every consequential action. No autonomous action on critical paths.
Immediate shutdown of AI systemsDerez, the kill-switch: sub-100ms, per-key and per-agent, behind 2FA.
Audit trail for AI-assisted workAppend-only, immutable, timestamped events. Examiner-ready export, 5-year retention.
Risk-management documentationSOPs codified as Workstreams and Skills: versioned, auditable, examinable.
Vendor oversightThe Grid centralizes every AI vendor connection, with unique keys per vendor and per agent.
Member-data handlingMember data hashed in audit logs. No persistent member data in model context.

The story, by who is asking

  • BSA Officer / CCO: every AI action is logged immutably with timestamp, actor, and context. Your team reviews everything before it is filed, and the kill-switch is on your dashboard, not ours.
  • CTO / CISO: each credit union's data lives in a dedicated instance. No shared infrastructure, no cross-tenant queries. Keys are unique per vendor and per agent, and all traffic flows through the Grid for full visibility.
  • CEO / Board: built for NCUA examination from day one. Human oversight is enforced by architecture, not policy, and you can shut down any AI agent in under 100 milliseconds.

Canonical reference

Real Estate Firms

Property and real-estate investment operations.

Philosophy of Work

How work should feel in an AI-native institution: trust earned in a loop, humans owning judgment, taste amplified.

On this page

Gated autonomyWhat NCUA requires, and how we answer itThe story, by who is askingCanonical reference