SentinelAI

Self-Hosting

Run the SentinelAI core on your own infrastructure

Self-Hosting

The SentinelAI core is open source and runs entirely on your own infrastructure. Data never leaves your network.

Why self-host

  • Data residency — prompts and responses stay inside your VPC
  • Regulatory control — no third-party processing of customer data
  • Cost predictability — no per-request cloud fees at scale
  • Full API surface — same detectors, same endpoints, same SDK

What you run

The self-hosted deployment is the same FastAPI service the managed cloud is built on:

ComponentPurpose
API service/api/analyze, /api/logs, /api/settings
Detector engineThe three parallel detectors
Policy engineTrust score thresholds and actions

Point the SDK at it

from sentinelai import SentinelAIClient

client = SentinelAIClient(
    api_key="sk_sentinel_...",
    base_url="http://localhost:8000",
)

Everything else — verify(), conversation context, corrected responses — works identically to the managed cloud.

What the managed cloud adds

The cloud is not a different product; it layers operations on top of the same core:

  • Real-time dashboard and alerting
  • Versioned settings history
  • Audit logs and team management
  • Managed uptime and scaling

Self-hosting is free and open source. Start with the core, move to the cloud when you want dashboards, alerting, and team features.

On this page