# Autopilot

Autopilot continuously adjusts a Security Group's challenge difficulty and enforcement policy based on the traffic Sentinel actually observes, instead of a fixed configuration you set once and forget. It's the recommended default for most deployments — strong security with minimal hands-on tuning.

## What it manages

Autopilot automatically activates and manages three things together:

- **Adaptive Proof-of-Work complexity** — raising or lowering PoW difficulty as traffic patterns shift.
- **[Adaptive Captcha](/docs/sentinel/features/adaptive-captcha/)** — deploying the Frictionless Captcha vs. Code Challenge split based on risk.
- **Default rate limiting** — a baseline limit of **50 requests per 10 minutes per EDK** ([Ephemeral Device Key](/docs/sentinel/security/privacy-protection/#ephemeral-device-keys-edk)), suitable for most deployments out of the box.

It also draws on [Threat Intelligence](/docs/sentinel/features/threat-intelligence/) to block requests from known-malicious IPs directly. Note that a request can be blocked before it ever reaches the rate limit above, if its overall threat score is already high enough on other grounds.

## Enabling it

Autopilot is **on by default** on new Security Groups. Toggle it from the Security Group's configuration in the admin interface — there's no separate env var to set. Turning it off hands full manual control back to you over PoW complexity, Adaptive Captcha, and rate limits via [Security Group rules](/docs/sentinel/configure/security-groups/#schema).

## Related

- [Adaptive Captcha](/docs/sentinel/features/adaptive-captcha/) — the mechanism Autopilot tunes.
- [Rate Limiters](/docs/sentinel/features/rate-limiters/) — the default rate limiting Autopilot applies, and how to configure limits manually.
- [Threat Intelligence](/docs/sentinel/features/threat-intelligence/) — the IP-reputation blocking Autopilot draws on.
- [Get Started](/docs/sentinel/get-started/) — Autopilot is on by default in a new deployment.
