Collection, detection, investigation and response in one deployment, with one data model underneath.
AIAn investigator, not a chatbot
A cyclic reasoning loop over your real telemetry with forensic tools it can call: process ancestry, network sessions, authentication history. It reaches a verdict with cited evidence, and says undetermined rather than guessing.
- Grounded in a 5,000-document security corpus, not the open web
- Runs on your hardware; no telemetry leaves the deployment
- Every stated fact carries the event ids it was drawn from
DETECTDetection that survives a real network
Sigma-native, with the full public corpus plus rules for Windows, Linux, Active Directory, firewalls, web servers and cloud identity. Coverage is computed at the engine with 24-hour hit counts, so a rule that never matched is a visible gap.
- 4,390 Sigma rules, validated on every change
- MITRE ATT&CK v19.1 map with per-technique 24h hit counts
- Correlation, sequence and stateful rules
RESPONDResponse with a hand on the brake
50 playbooks covering the attack classes a SOC actually sees. Containment sits behind an approval gate that genuinely blocks: an approval performs the action, a rejection means nothing happened.
- Host isolation, account disable, IP blocking, token revocation
- Every step recorded against the case with who approved it
- Dry-run mode so a playbook can be proven before it is armed
OPERATEBuilt to be run by a team
Cases, evidence custody, sign-off, and an audit log that is cryptographically signed. Incidents close through an independent approver, so resolved means somebody accountable said so.
- RBAC with viewer / analyst / admin separation and mandatory MFA
- Evidence attached to the case and carried into the report
- Tickets are the durable record; they outlive event retention