Description
In this short video we demonstrate how personally identifiable information can be prevented from leaking to LLMs using a data contract enforced by Gable.
Intro to Gable
Quick orientation to Gable’s shift-left data governance platform: how it catches application changes in the CI/CD workflow to protect data trust before anything reaches production.
Demo Overview
A brief introduction to this demo which uses the Open Bank project to demonstrate how to guard against data generating code leaking personal identifying information to the Open AI APIs.
Open Bank Application
Sets the scenario with a mock Open Banking microservice that emits customer transactions, giving real-world context for financial-grade data privacy and compliance challenges.
Opey / Open Bank APIs
Shows “Opey,” Gable’s reference implementation of PSD2-style Open Banking APIs, and how Gable instruments them to track contracts and sensitive fields across requests.
An Example of a PII Leaking Pull Request
Highlights a developer PR that unintentionally exposes personally identifiable information, creating a concrete use case for automated policy checks.
PII Detection
Demonstrates Gable’s pattern- and type-aware scanners flagging that PR in real time, classifying email, account, and tax-ID fields as regulated PII.
CI/CD Enforcement
Shows the failed GitHub Actions run where Gable blocks the merge, surfaces the exact offending lines, and gives actionable remediation guidance—all inside the pipeline.
Static Code Analysis
Covers Gable’s AST-level inspection that maps variable names, data types, and serialization logic, enabling early detection of breaking data-contract changes as well as to generate draft data contracts.
Downstream Subscriber Notification
Illustrates automatic Slack and webhook alerts sent to data consumers—dashboards, ML models, and data warehouses—so teams know contract changes before incidents occur.
Data Lineage
Concludes with a live lineage graph tracing the affected fields to the downstream data assets that depend on them.