Sentinel AI

Sentinel AI is an enterprise AI operations command center designed to investigate data incidents, understand their business impact, generate remediation, and produce executive reports.

Modern data platforms generate enormous amounts of information, but when something breaks, teams need more than an error message. They need to understand what changed, what depends on the affected data, who owns it, which business processes are impacted, and what should happen next.

Sentinel AI combines Gemini-powered reasoning with structured DataHub-compatible enterprise context, including metadata, schemas, ownership, governance, assertions, data contracts, and lineage.

The platform provides a six-step guided investigation: incident detection, enterprise context retrieval, lineage discovery, root-cause analysis, remediation generation, and executive reporting.

In the demonstration scenario, Sentinel investigates a Revenue Analytics pipeline failure, traces its downstream impact across pipelines, dashboards, and ML models, identifies unexpected NULL values in a payment currency field, generates an SQL/dbt remediation patch, and produces an executive post-mortem.

The application also includes an interactive lineage graph, enterprise metadata catalog, specialized agent workflow, incident manager, DataHub context panels, and animated investigation experience.

Gemini provides real AI-powered analysis and code generation through a secure server-side integration. The current hackathon release uses deterministic DataHub-compatible context rather than a live DataHub MCP Server, allowing the demonstration to remain reliable and reproducible while maintaining a clear path toward live MCP integration.

Sentinel AI demonstrates how enterprise context can transform AI agents from systems that simply answer questions into systems that can investigate, understand relationships, generate solutions, and communicate their findings.

AI agents are only as effective as the context they have. Sentinel AI turns enterprise context into autonomous action.

Built With

• datahub
• express.js
• node.js
• react
• typescript
• vite