Technical Overview

This page is for security, data, and platform teams. It outlines how Rivlo integrates with your environment, how execution works, and what controls are available.

Back to Home Request a Demo

How Rivlo Works.

Rivlo is a desktop-first analytics workspace. It connects to your databases and help generate SQL using schema-only metedata. No row-level data is ever sent to the model. The goal is controlled, reviewable analysis - without creating more operational risk.

Schema Only

The model receives table names, column names, and relationships to understand structure.

No Data Leaves

Row-level data, query results, and database contents are never sent to the model endpoint.

Local Execution

Queries execute against your databases within your environment. Results are returned directly to the local Rivlo client.

Architecture

Rivlo is designed to keep execution close to your data and make the workflow easy to inspect. The sections below describe the common control points.

Execution Path

Queries execute against your configured databases using the credentials and permissions you provide.

Review Point

SQL is visible and reviewable before execution, making it easier to enforce internal standards.

Reports Storage

Reports store the SQL and a snapshot of results so teams can reference what was shared and re-run intentionally.

High-Level Flow

Your Environment

Databases, permissions, and network controls stay under your policy.

Rivlo Desktop

• Connect to configured databases
• Draft and review SQL
• Execute and save Reports

Model Endpoint

Rivlo hosts the LLM through AWS Bedrock using Claude. Only schema context is sent to the model.

Implementation Note

The exact boundary and controls depend on your environment. We validate the deployment model during evaluation.

Security Model

Rivlo is built around clear control points: who can connect, what can be queried, and what gets saved. Specific controls vary by environment and policy.

Reviewable SQL

SQL is visible and can be inspected before execution. This supports internal review and training workflows.

Least Privilege

Rivlo uses the database permissions you configure. You decide what users can access and run.

Reports Retention

Reports store SQL and a results snapshot for repeatability. Retention and storage location can be aligned to policy.

Security questions we expect

Where does execution happen?

Against your configured databases using the credentials and permissions you provide.

What model do you use?

Rivlo uses Claude through AWS Bedrock for managed model inference.

What is sent to the model?

Only schema context is sent to the model. No row-level data, query results, or database contents are transmitted to or stored by Rivlo.

What gets saved in Reports?

Reports are designed to store the SQL and a results snapshot for consistent sharing and repeatability. Retention policy is typically set during evaluation.

Note

We’ll align the final controls and configuration with your compliance and IT requirements.

Technical FAQ

Questions from security, data, and platform teams.

Do you support SSO?

SSO expectations vary by customer. We’ll align authentication and access controls during evaluation.

Can we restrict query patterns?

Most teams enforce restrictions through database permissions and review workflows. We’ll validate your constraints during demo/eval.

How do you handle auditability?

The core workflow is designed to be reviewable and repeatable. Reporting and retention policies are configured to match internal needs.

What’s the best way to evaluate Rivlo?

Start with a sandbox schema and a small set of representative questions. We’ll propose a safe evaluation plan.

Request a Demo

Or go back to the home page.