---
title: "How Rex uses and secures AI"
description: "Rex uses large language models to run accounts receivable workflows — chasing invoices, reconciling payments, managing disputes, communicating with customers — with tenant isolation and auditable traces."
canonical: https://www.rex.inc/ai-security
last-updated: 2026-08-21
---

# How Rex uses and secures AI

Rex uses large language models to run accounts receivable workflows — chasing invoices, reconciling payments, managing disputes, communicating with customers — with tenant isolation and auditable traces.

## Architecture

Rex uses Anthropic's Claude models for all LLM inference, accessed via AWS Bedrock. Rex does not train, fine-tune, or host its own models. All inference runs server-side. End users never interact with the model directly. Every request is scoped to a single tenant.

## Data

Customer data is used at inference time to perform work within that tenant. Operational data is retained for the duration of the contract. Tenant data is not used to train or fine-tune models.

## Further reading

- Full page: [How Rex uses and secures AI](https://www.rex.inc/ai-security)
- [Trust Center](https://trust.rex.inc)
- [Privacy policy](https://www.rex.inc/privacy-policy)
- [Contact](https://www.rex.inc/contact)