Services
Security services, delivered on AGT.
Incident response, managed detection, an agentic SOC, training and adversary assessments. Run by practitioners with decades in the field, on the blue/red platform we build ourselves.
Service lines
From the first alert to a stronger defense
Five services, one platform. Our practitioners run each engagement on AGT, so the evidence, the detections and the lessons all end up in one place you keep.
01 Incident Response
From the first alert to a verified eviction.
When something is wrong, the first hours decide how bad it gets. We put the AGT sensor on the affected Windows and Linux hosts, pull the artifacts that answer what ran, when, and from where, and build the timeline with you on the call.
Containment comes first and is reversible. Eviction happens all at once, so the actor can’t watch you rotate credentials. Every step we take on your hosts is recorded with its reason.
You receive an incident report with every affected host, account and indicator, a remediation plan built from the evidence, and a STIX 2.1 bundle for Intel Consortium or your threat-intel platform.
- 02:14:07
WS-FIN-07
outlook.exe→powershell.exe -enc …critical T1059.001 - 02:14:31
WS-FIN-07
Run key added:
OneDriveSynchigh T1547.001 - 02:19:52
WS-FIN-07
lsass.exeopened byrundll32.execritical T1003.001 - 02:41:10
SRV-FILE-02
SMB logon as
svc_backuphigh T1021.002
- Contain
- Evict
- Recover
- Harden
- Verify
02 Managed Security
Detection that is tuned, swept and measured.
Most detection programs fail quietly: rules that never fire, sensors that stopped sending, and alerts nobody had time to read. We run AGT across your Windows and Linux estate, containers included, and measure all three.
Our analysts triage the alert queue and escalate with the evidence attached. Scheduled sweeps check every host for indicators and YARA signatures, and hardening reviews close the gaps detection can’t.
You receive triaged alerts, fleet-wide sweep results, and a running account of where your coverage is thin.
- critical LSASS access by unsigned process 1 host · 4m ago
- high Encoded PowerShell from an Office child 3 hosts · 12m ago
- high Defender exclusion added 1 host · 2h ago
- medium New service installed 7 hosts · 5h ago
- 123ATT&CK-mapped rules
- 9never fired in 30 days
- 02:00nightly YARA sweep
03 Agentic SOC
Machines work the leads. People make the calls.
An agentic SOC is not a chatbot in front of your alert queue. In AGT, cases move through pipelines: deterministic tools collect and sweep, models reason over what they return, and a gate stops the pipeline wherever a person should decide.
We design the pipelines with you and set the gates where your risk tolerance puts them: new detection rules, case conclusions, anything else. The model is your choice, self-hosted or from Anthropic, OpenAI or Google.
You receive cases worked end to end with a trail you can review, and detection rules drafted from the evidence and held for your approval.
- Lead triage 7 leads in, 3 kept done
- Indicator sweep 214 hosts swept, 2 hits done
- Correlate linked WS-FIN-07 → SRV-FILE-02 done
-
Draft detection rule
approve
smb-logon-svc_backup? waiting on you - Conclude and report queued behind the gate queued
04 Security Training
Practice against a real adversary, scored on what happened.
The best training is the real thing, run somewhere it can’t hurt you. We run blue, red and purple courses on our own Active Directory range, with AGT on both sides: the blue team works the sensor and the alert queue, the red team works the engagement, and the console keeps score.
Trainees see only their own side; the instructor sees both and leads the debrief. Scores come from a fixed table, so a result means the same thing on the last day of the course as on the first.
You receive per-trainee progress against the course objectives, a match report with every event that scored, and a replayable record of the exercise.
- +150 Eviction confirmed on WS-02
- +120 Domain admin credential
- −20 False positive raised
- +80 Caught: LSASS access
- +60 Foothold on WS-02
05 Security Assessments
Red teaming and network penetration testing that leave you stronger.
An assessment is only worth what changes because of it. We agree the scope and rules of engagement in writing before anything runs, then test the networks you authorize and pursue the objectives you set, across Windows, Linux and Active Directory.
AGT keeps the engagement inside that scope and records every credential, foothold and cleanup step. If your hosts run the AGT sensor, the same console scores the engagement against your detections: what was caught, what was missed, and what to change.
You receive a debrief of objectives, findings and cleanup, and prioritized remediation tied to the finding behind it.
- Credential dumping T1003 caught
- Lateral movement over SMB T1021.002 missed
- Scheduled-task persistence T1053.005 caught
- Kerberoasting T1558.003 missed
- Out-of-scope host 10.30.4.12 halted
The platform
AGT: one console for blue and red
Every service runs on AGT, our security platform: an endpoint sensor and detection console for defenders and engagement tooling for the red side, in one self-hosted console. It is still in development, and our services are how it reaches customers today.
- Your data stays put AGT is self-hosted and nothing phones home. We can run it inside your environment, so telemetry and evidence never leave it.
- Blue and red, scored together Both halves share one console, so every red action can be matched to the detection that caught it, or recorded as a gap.
- Feeds Intel Consortium Cases export as STIX 2.1, so indicators from an incident land in your Intel Consortium graph, or any platform that reads STIX.
Need help now, or planning ahead?
Tell us what you are dealing with. We will scope it with you and tell you plainly what we can and can't do.