Platform
One system for the whole alert lifecycle.
Solomon ingests your alerts and telemetry, maps your environment as it actually runs, and puts agents to work on your alerts, with a human approving response actions by default.
01
Ingest
Connect identity, endpoint, cloud, network and SIEM sources. Solomon resolves the same user, host or service across tools into one entity.
- identity providersvc-deploy@northwind-lab.example
- cloud auditarn:aws:iam::111122223333:role/svc-deploy
- endpointNWL\svc-deploy
- ticketingsvc-deploy (service account)
svc-deploy
- aliases
- 4
- first seen
- 2026-01-15
- facts
- 1,247
02
Map
A live runtime map of workloads, services, identities and data stores, with the traffic between them. Scrub back in time to see the environment as it was during an incident.
03
Investigate
Before running a single query, Solomon decides which sources an alert needs and in what order. The plan is fixed and inspectable, so you see what it will check before it checks it.
- 1Who signed in, from where?idp.signinsfree1u
- 2Is this hour normal for the account?memory.baselinefree1u
- 3What did the session touch?cloud.auditinternal2u
- 4Is the source address known-bad?intel.lookuppaid5u
04
Decide
Verdict, confidence and a recommended action, with every claim linked to its evidence. Approve, modify or reject.
A session token for svc-deploy was approved after nine denied MFA pushes, used at 02:14 UTC, outside the account’s usual hours, and then listed storage buckets for the first time in 30 days.
Recommended Revoke all sessions for svc-deploy
blast radius: 1 identity · rollback: re-issue on next sign-in
05
Learn
Verdicts and analyst corrections are written back to memory and to detection coverage, so the next alert starts warmer.
Teach Solomon
backup-runner signs in at 02:00 UTC every night to copy snapshots. That’s expected, not anomalous.
- Applies to
- this entity · backup-runner
- Expires
- in 90 days
- Would have changed
- 3 past cases, all from inconclusive to benign
Governance
A deterministic policy layer sits between what an agent concludes and what it is allowed to do. Response actions need approval by default, every agent action is logged, and an emergency stop puts a tenant in observe-only mode.



