HUNT ACTIVE
Part of the Citadel Cyber Defense Platform · Brindwell & Partners

Your SIEM waits for alerts. Vanguard goes hunting

57% of breaches are discovered by outsiders — not your SOC, not your SIEM, not your EDR. The attackers who evade automated detection are the ones who cause catastrophic damage. They use legitimate tools. They move slowly. They blend in. They are patient. Citadel Vanguard hunts them — proactively, continuously, relentlessly — using hypothesis-driven investigation, behavioral analytics, and AI-augmented telemetry to find what your automated defenses were designed to miss.

VANGUARD HUNT CAMPAIGN
HYPOTHESIS ACTIVE
HYPO-0147HUNT "APT group is using WMI for lateral movement in our environment"
QUERYSCAN Searching 14,000 endpoints for WMI process creation anomalies
RESULTANOMALY 3 endpoints show WMI activity outside business hours — WKSTN-118, SRV-DC02, SRV-FILE04
ENRICHCONTEXT WMI source: service account svc_backup — no legitimate WMI usage in baseline
VERDICTTHREAT Compromised service account — lateral movement confirmed — ESCALATING
241→24h
Dwell time reduction with mature hunting
57%
Of breaches found by outsiders (not your SOC)
More threats found vs. detection-only programs
$4.4M
Average cost of a breach (IBM 2025)
The Detection Ceiling

Your detection stack has a ceiling. It will always have a ceiling. Because the most dangerous adversaries design their operations specifically to stay below it. They use PowerShell, not malware. They use RDP, not exploits. They use your own service accounts, not their own credentials. They move at human speed — one step per day, one lateral movement per week — staying below the thresholds your detection rules were tuned to catch. These are the adversaries that cause $4.4 million breaches. And your SIEM will never alert on them.

Vanguard exists in the space above the detection ceiling — where human expertise, hypothesis-driven investigation, and AI-augmented analytics converge to find the threats that automation was never designed to catch. Every hunt begins with a question: "What would this specific adversary do in our specific environment?" Then the hunter goes looking for the answer — armed with behavioral baselines, cross-domain telemetry, threat intelligence, and ML anomaly detection that surfaces the subtle signals hidden in petabytes of normal activity.

Hunting Engines

Eight engines. Human expertise amplified by machine intelligence.

Vanguard provides the platform, the data, and the AI — your hunters provide the hypotheses, the intuition, and the judgment.

ENGINE 01
Hypothesis-Driven Hunt Campaigns
Structured hunt workflows — from intelligence-driven hypothesis generation through data investigation, validation, and detection rule creation.
AI suggests hypotheses based on your threat profile, industry, and ATT&CK gaps

Every hunt starts with a question — a falsifiable hypothesis about adversary behavior. "An attacker has compromised a service account and is using WMI for lateral movement." "A threat actor is staging data in a cloud storage bucket before exfiltration." "An insider is using legitimate tools to access data outside their role." Vanguard structures the entire hunt lifecycle: hypothesis generation (AI-suggested based on your industry, threat profile, and ATT&CK coverage gaps), data scoping, query execution across all telemetry sources, anomaly analysis, validation, and — critically — conversion of validated findings into automated detection rules so the same threat is caught automatically next time.

Capabilities
AI
Hypothesis suggestions based on threat profile and detection gaps
100%
Hunt findings converted to detection rules via flywheel
ENGINE 02
Living-off-the-Land Detection
Specialized analytics for detecting adversaries using legitimate tools — PowerShell, WMI, PsExec, RDP, certutil, and LOLBins.
Detects LOTL techniques that generate zero malware signatures

The most sophisticated adversaries never install malware — they use the tools already present on every Windows system. PowerShell for command execution. WMI for lateral movement. Certutil for file download. PsExec for remote execution. RDP for interactive access. These "living-off-the-land binaries" (LOLBins) are legitimate administration tools — and they generate zero malware signatures. Vanguard's LOTL detection engine baselines normal administrative tool usage for every user, every service account, and every system — then hunts for deviations that indicate adversary abuse: PowerShell execution by users who never use it, WMI connections at unusual hours, PsExec from workstations that should never initiate remote administration.

Capabilities
40+
LOLBin tools baselined and monitored per environment
User
Per-user, per-system behavioral baselines for legitimate tool usage
ENGINE 03
Adversary TTP Intelligence
Real-time threat intelligence mapped to MITRE ATT&CK — prioritized by relevance to your industry, geography, and technology stack.
Tracks 200+ threat groups and maps their TTPs to your environment

Effective threat hunting requires knowing who is likely to target you — and how they operate. Vanguard's adversary intelligence engine maintains continuously updated profiles of 200+ documented threat groups, mapping their preferred tactics, techniques, and procedures to the MITRE ATT&CK framework. The system prioritizes threat groups by relevance to your industry vertical, geographic presence, technology stack, and known targeting patterns — so your hunters focus on the adversaries most likely to be in your environment, not the ones making headlines in someone else's industry.

Capabilities
200+
Threat group profiles maintained with TTP mapping
Auto
Industry-specific threat prioritization based on your profile
ENGINE 04
Behavioral Anomaly Analytics
ML-driven behavioral baselines for every user, device, and service — surfacing subtle deviations that indicate compromise or insider threat.
Detects "slow and low" adversary movement invisible to threshold-based alerts

The adversaries who cause the most damage are the ones who move slowly — one lateral movement per day, one new access per week, always staying below detection thresholds. Vanguard's behavioral analytics engine builds dynamic baselines for every entity in your environment: users (login times, access patterns, data volumes), devices (process execution, network connections, service startups), and services (API call patterns, data flow volumes, authentication behavior). The engine surfaces anomalies that individually seem benign but collectively indicate compromise — a login from a new location + access to an unusual file share + a data transfer slightly above average = a pattern that deserves investigation.

Capabilities
Entity
Per-user, per-device, per-service behavioral baselines
Multi
Multi-signal correlation detects compound anomalies from benign components
ENGINE 05
Cross-Domain Hunt Telemetry
Unified hunt data lake spanning endpoint, network, cloud, identity, email, and SaaS — queryable with natural language and structured queries.
Hunt across every data domain from a single query interface

Hunters can only find what they can see — and most hunting programs are limited by fragmented data access across siloed tools. Vanguard provides a unified hunt data lake that aggregates telemetry from endpoint (process, file, registry, network), network (flow, PCAP, DNS), cloud (audit logs, IAM, storage), identity (AD, Entra, Okta), email (gateway, inbox), and SaaS (O365, Salesforce, Slack) into a single, high-performance query interface. Hunters can search across all domains simultaneously using natural language ("Show me all PowerShell execution by service accounts in the last 30 days") or structured queries (KQL, SPL, Sigma) — eliminating the tool-switching that consumes 40% of a hunter's time in fragmented environments.

Capabilities
6+
Telemetry domains unified in single hunt data lake
NL
Natural language and structured queries across all data
ENGINE 06
Hunt-to-Detection Flywheel
Every validated hunt finding automatically generates a detection rule — so your automated layer catches the same behavior next time.
Converts hunt discoveries into Sigma/YARA/KQL detection rules automatically

A hunt without a detection flywheel is a missed opportunity. Every time a hunter finds a threat that automated detection missed, the natural question is: "Why didn't our SIEM catch this?" Vanguard's flywheel engine answers that question and solves it. When a hunt validates a finding, the system automatically generates a detection rule in the appropriate format (Sigma, YARA, KQL, SPL), tests the rule against historical data to measure false positive rate, and deploys the validated rule to Citadel's SIEM. Over time, the flywheel continuously expands automated coverage based on real-world hunt findings — so each hunt makes the next one less necessary for that specific technique.

Capabilities
Auto
Sigma/YARA/KQL rule generation from validated hunt findings
FP
Automated false positive testing against 90 days of historical data
ENGINE 07
Dark Web & External Intelligence
Monitors dark web forums, paste sites, criminal marketplaces, and telegram channels for your organization's credentials, data, and mentions.
Early warning of credential compromise and data exposure before attack execution

The most valuable threat intelligence often comes from outside your perimeter — from the criminal underground where stolen credentials are sold, where initial access brokers auction network footholds, and where threat actors discuss targeting your industry. Vanguard monitors dark web forums, paste sites, criminal marketplaces, Telegram channels, and underground access broker sites for your organization's credentials, domain mentions, data samples, and infrastructure references. When your data appears in the underground, Vanguard generates an immediate hunt — searching your environment for evidence that the compromised credentials have been used or that the advertised access point has been exploited.

Capabilities
24/7
Continuous dark web and underground monitoring
Auto
Immediate hunt triggered when organizational data appears underground
ENGINE 08
Managed Hunt Operations
24/7 threat hunting by Citadel's elite hunt team — for organizations that need hunting capability without building an in-house program.
Elite hunters with an average 12 years of adversary experience per analyst

Threat hunting requires a rare combination of adversary mindset, deep technical expertise, and investigative intuition — talent that most organizations cannot recruit or retain. Vanguard's managed hunt service provides dedicated Citadel hunters who conduct continuous hunt campaigns against your environment using your telemetry, your threat profile, and your risk priorities. Each hunter averages 12 years of adversary-focused experience, including backgrounds in national intelligence, law enforcement cyber units, and enterprise red teams. Findings are delivered with full context, MITRE ATT&CK mapping, and detection rules — and integrated directly into your Citadel SIEM and SOAR workflows.

Capabilities
12yr
Average adversary experience per managed hunt analyst
24/7
Continuous managed hunting with monthly strategic briefings
Hunt Results
Fortune 500 Energy Company

Hunt discovered 18-month dormant APT implant in SCADA adjacent network

Vanguard's managed hunt team executed a hypothesis-driven campaign targeting OT/IT boundary reconnaissance. A behavioral anomaly in DNS query patterns from a historian server — subtle frequency changes invisible to threshold-based detection — led to the discovery of a dormant implant that had been present for 18 months. The implant was a custom backdoor communicating via DNS tunneling at intervals designed to mimic legitimate NTP traffic. Attribution analysis linked it to a documented nation-state threat group targeting energy infrastructure. The implant was eradicated without operational disruption. Post-hunt detection rules now monitor for DNS tunneling across all OT-adjacent systems continuously.
18mo
Implant dwell time
DNS
Tunneling detected
0
Operational disruption
Global Insurance Group — Insider Threat Hunt

Behavioral analytics uncovered systematic data staging by departing executive

A hypothesis hunt focused on data access patterns by users in their notice period uncovered a departing VP who had been systematically accessing client portfolio data outside their normal role scope for 6 weeks. Vanguard's behavioral baseline showed the executive's data access volume had increased 340% since giving notice — spread across small increments designed to stay below DLP thresholds. Cross-domain telemetry revealed the data was being staged in a personal OneDrive folder via browser upload, bypassing the corporate DLP that only monitored email and USB. The investigation prevented the exfiltration of $2.3B in client portfolio data to a competing firm.
340%
Access anomaly
$2.3B
Data protected
6wk
Staging period caught
Healthcare System — Dark Web Triggered Hunt

Credential sale on dark web triggered hunt that found active compromise

Vanguard's dark web monitoring detected credentials for 47 employees of a healthcare system being sold on a Russian-language criminal forum. The automated hunt triggered by this finding searched authentication logs across all systems for those credentials — and found that 3 of the 47 had already been used by an external actor to access the VPN, email, and an internal SharePoint site containing PHI. The compromise had been active for 11 days with no SIEM alert generated because the attacker used valid credentials during business hours from a US-based VPN exit node. All 47 passwords were reset, the 3 compromised accounts were forensically investigated, and the external access was terminated. No PHI was confirmed exfiltrated.
47
Credentials found
3
Active compromises
11d
Dwell time ended
Hunter Voices

The DNS tunneling implant had been in our environment for eighteen months. Eighteen months. Our SIEM processed the DNS traffic every day and never alerted — because the traffic volume was designed to look like NTP updates. Our EDR saw a legitimate Windows service making DNS queries and flagged nothing. Our network monitoring saw encrypted traffic to a known DNS resolver and ignored it. A Vanguard hunter noticed that the query frequency deviated from actual NTP patterns by 0.3 standard deviations. That deviation — invisible to every automated system we own — was the thread that unraveled an 18-month nation-state operation in our SCADA network.

VP of Cybersecurity
Critical Infrastructure Security
Fortune 500 Energy Company

The detection flywheel is the most underappreciated feature. Every hunt that finds something produces a detection rule. Every detection rule means the automated layer catches that behavior next time. After six months of Vanguard hunts, our automated detection coverage expanded by 34 new rules — each one based on a real threat found in our real environment by a real hunter. Not theoretical rules from a vendor template. Rules born from actual adversary behavior. That is how you turn proactive hunting into permanent defensive improvement.

Director of Detection Engineering
SOC Architecture & Strategy
Global Financial Services Firm

We discovered our departing VP was stealing $2.3 billion in client portfolio data — and our DLP never saw it. The data was leaving via browser upload to a personal OneDrive account, in increments small enough to stay below every threshold we had set. Without the behavioral baseline that showed a 340% increase in data access since the notice period, and the cross-domain telemetry that tracked the browser uploads our DLP didn't monitor, that data would have walked out the door. Vanguard did not just find a threat. It prevented a business catastrophe.

Chief Information Security Officer
Cybersecurity & Risk
Global Insurance Group
241→24h
Dwell time reduction
200+
Threat groups tracked
More threats found
34
New rules per 6mo (avg)
Find What Your SIEM Cannot

The adversary is already inside. Start hunting.

Deploy Vanguard — as a platform for your hunt team, or as a managed service with Citadel's elite hunters operating in your environment.

Or request a complimentary hunt assessment at vanguard@brindwell.com