Every alert, investigated
the way your best analyst would
Carpent's agents take the alerts from your EDR, SIEM, email and identity tools, rebuild what happened and act with the autonomy you choose for each action and each asset.
✓Included in the complete platform and available on its own
Cases
Last 24 hours
Awaiting approval
2
Investigating
5
Contained by agents
11
Alerts dismissed
1.284
Isolate SRV-APP-07
Proposed by Agent-07 · «Production servers» policy requires approval
Agent-07 · razonamiento
j.ruiz signed in through vpn.acme.example without MFA with a password found in a Lumma log. wmiexec and an LSASS dump ran on SRV-APP-07. Isolating it cuts the path to DC-ZGZ-01.
Already done by the agents
✓ j.ruiz VPN session cut
✓ 198.51.100.12 blocked at the perimeter
✓ j.ruiz password reset
Impact of the action
- Affected applicationCompras ERP
- Active users14
- ReversibleYes
- On callSara Vidal
- 00:00.4EDRpowershell.exe -enc spawned by winword.exe on FIN-LT-0231
- 00:03.1Entra IDm.ortega signed in from 185.220.101.47 12 min ago
- 00:05.8Threat intel185.220.101.47 · Tor exit node · 14 reports
- 00:09.2EmailAttachment Invoice_0923.docm received at 02:51
- 00:14.6NetworkBeacon to cdn-sync[.]net:443 every 60 s
- 00:21.0AgentChain phishing → macro → C2. Containment advised.
Entities
FIN-LT-0231
m.ortega
185.220.101.47
cdn-sync[.]net
MITRE ATT&CK
T1566.001
T1204.002
T1059.001
T1071.001
Verdict · Agent-07
True positive
Confidence 0.97
Every alert reaches an agent that investigates it the way your best analyst would. Noise closes itself, with its reasoning on record. What is real gets contained within the limits you set. Your team only gets what needs human judgement.
What it does
Full investigation
Every alert is enriched with asset, user and threat intelligence context, and ends in a verdict with its reasoning.
Containment within limits
Isolate a device, block an account or pull an email. Each action has its own policy, scope and approval mode.
Escalation with the case built
Whatever needs human judgement arrives as a case with the chain rebuilt, the evidence and the proposed action.
Everything on record
Every query, decision and action an agent takes goes into an audit log that cannot be altered.
In the console
Platform screens with data from Acme, the fictional company in the live demo.
Cases
Last 24 hours
Awaiting approval
2
Investigating
5
Contained by agents
11
Alerts dismissed
1.284
Isolate SRV-APP-07
Proposed by Agent-07 · «Production servers» policy requires approval
Agent-07 · razonamiento
j.ruiz signed in through vpn.acme.example without MFA with a password found in a Lumma log. wmiexec and an LSASS dump ran on SRV-APP-07. Isolating it cuts the path to DC-ZGZ-01.
Already done by the agents
✓ j.ruiz VPN session cut
✓ 198.51.100.12 blocked at the perimeter
✓ j.ruiz password reset
Impact of the action
- Affected applicationCompras ERP
- Active users14
- ReversibleYes
- On callSara Vidal
How it works
01Ingest
Alerts arrive from your tools by API, webhook or syslog and are grouped when they describe the same incident.
- ✓Asset and user context
- ✓Mapping to MITRE ATT&CK
- ✓Verdict with its reasoning
02Investigate and decide
The agents query every source, map the activity to MITRE ATT&CK and decide whether it is a false positive or a threat.
- FIN-LT-0231Isolated
- m.ortegaSession revoked
- CAS-2317Your approval
03Contain or escalate
If the policy allows it, they contain straight away. If not, they ask the on-call person for approval with everything ready.
Connected to the rest of the platform
Assets and vulnerabilities from Exposure management weigh on each alert's priority, and indicators from Threat intelligence feed every investigation.
See the complete platform [ → ]Other modules

