Governor Server Sentinel
A Linux service that gathers evidence from servers, analyses incidents with an advisory AI and sends alerts. Memory is kept strictly separate for each client, and the operator chooses exactly how much the service may do.
THE PRODUCT, EXPLAINED
See the warning. Choose the response.
Start with the short film. Then see how it works in the full demonstration.
Follow server trouble and security warning signs into a clear, controlled response.
Press Play. CC adds captions. Use full screen for the clearest view.
Illustrative scenarios. * Extended features in final R&D.
Evidence first, then analysis, then an alert
When something goes wrong on a server, the operator's first need is evidence: what the host was doing, what changed and what the signals say. Governor Server Sentinel is a Linux service that gathers that evidence from the servers it watches, analyses the incident with an advisory AI and sends the operator an alert with the analysis attached.
The service is live in advisory mode across multiple internal and external pilots. Sentinel is designed around three response tiers, from recommendation only to a standing policy with an exact scope and an expiry, so an operator decides how far the service may go. Memory is kept strictly separate for each client, so one operator can serve several organizations cleanly.
Linux service
Runs on the fleet it watches and gathers evidence from the servers directly.
Advisory AI
Analyses the gathered evidence and recommends; the analysis travels with the alert.
Per-client memory
Memory is kept strictly separate for each client. Nothing learned about one organization leaks into another.
From signal to recorded response
The same sequence runs in every tier; what differs is how much the service may do at the last step.
-
1
Gather evidence
Sentinel collects evidence from the servers it watches: the state of the host and the signals around the incident.
-
2
Analyse with an advisory AI
The evidence is analysed by an advisory AI, which produces an assessment of the incident and a recommended response.
-
3
Alert the operator
An alert is sent to the operator with the evidence and the analysis, so the recommendation can be judged on what was actually observed.
-
4
Respond within the chosen tier
In the advisory tier the operator acts. In the managed tier each action needs approval. In the autonomous tier a standing policy with an exact scope and an expiry lets Sentinel respond within those bounds.
Three response tiers
Sentinel is designed around three tiers, so an operator decides how far the service may go. Evidence comes first in every tier.
Advisory
Sentinel gathers the evidence, analyses the incident and recommends. The operator acts.
- Evidence collected from the servers
- Incident analysis by an advisory AI
- Alerts sent to the operator
- The tier in internal pilot use today, live across multiple internal and external pilots
Managed
Sentinel proposes a response, and each action needs the operator's approval before it runs.
- One approval per action
- Recommended response with its evidence
- Operator keeps the final decision
Autonomous
A standing policy with an exact scope and an expiry lets Sentinel respond within those bounds.
- Exact scope, explicit expiry
- Demonstrated on isolated test systems, with a Linux service restart as the reference action
- Same evidence-first discipline as the other tiers
| Tier | What Sentinel does | Who acts | Bounds | Where exercised |
|---|---|---|---|---|
| Advisory | Gathers evidence, analyses, recommends, alerts | The operator | Recommendation only | Live across multiple internal and external pilots |
| Managed | Proposes a response with its evidence | The operator approves each action | One approval per action | Designed tier |
| Autonomous | Responds within a standing policy | Sentinel, within the policy | Exact scope and an expiry | Isolated test systems: Linux service restart |
A monitor built for governed fleets
Server Sentinel and Server Management are designed to work alongside Governor AI as the operator's additional safety layer around governed automation.
Per-client separation
Memory is kept strictly separate for each client, so one operator can serve several organizations cleanly.
Evidence before action
Every tier starts from collected evidence and recorded analysis, the same discipline Governor AI applies to every decision.
Authority the operator sets
The three tiers mirror the Governor principle that authority is explicit: recommendation, approval per action, or a standing policy with exact bounds.
Pairs with Server Management
Sentinel supplies the evidence and the analysis; the console gives the operator the fleet-wide wall and the approvals view.
Standalone program
A reusable Governor program with one authoritative source; client-specific configuration and state kept separate.
Internal pilot
Like every built Governor product, Sentinel runs as an internal pilot, live in advisory mode across multiple internal and external pilots.
Two products share a name. They are not the same thing. Governor Server Sentinel is this infrastructure monitor for servers and fleets. Governor Technologies also runs a separate security-research effort that uses the Sentinel name; it lives on the Security Research page and is not part of this product.
Governor Server Sentinel at a glance
Talk to us about your fleet
Request a briefing on Governor Server Sentinel and the response tier that fits your operations.
Request a Briefing