Investigation Pipeline

The AI Security Analyst — How ARIA Investigates an Alert

ARIA's AI analyst does not score alerts. It investigates them: gathering evidence from multiple sources, producing a verdict where each claim cites its source, and routing the result through independent verification before any automated action.

Step One — Deterministic Enrichment Before AI

The most common failure mode in AI triage systems is asking the model to do two things at once: gather evidence and reason about it. When a model is given a raw alert and asked "is this a threat?", it is guessing based on what it has seen in training — not investigating based on actual evidence from this environment. ARIA separates these steps.

When an alert arrives, a deterministic enrichment pipeline runs first. This is not AI — it is a sequence of structured queries with specific outputs. The MITRE ATT&CK technique is taken directly from the Wazuh sensor's tag, not inferred. Threat intelligence lookups run against OTX, CISA KEV, and other configured feeds, and return specific results: whether the indicator appears, with what confidence and context. Alert history for the involved entity — recent events, similar patterns — is pulled from the alert database. For cloud alerts, identity context is collected: the user's MFA status, assigned roles, and recent login history.

Only when this enrichment is complete does the alert move to the AI triage stage. The model receives not just the raw alert but the full enrichment package — which changes what the model can say and what it can be held to.

Step Two — Structured Verdict with Cited Evidence

The triage LLM receives the enriched alert and produces a structured verdict. The structure has a specific requirement: each factual claim in the verdict must cite the enrichment source that supports it. If the verdict states that an IP address has a known malicious reputation, that claim must be grounded in the threat intel lookup result — not inferred from training data about that IP.

This citation requirement serves two purposes. First, it makes the verdict verifiable: a human reviewing the alert can trace each claim to its source. Second, it constrains what the model can say: claims that would require the model to invent evidence are structurally disallowed. A verdict that tries to claim "this pattern is typical of ransomware staging" without a supporting enrichment result does not pass validation.

The structured output also captures uncertainty explicitly. When the evidence is genuinely ambiguous — an indicator that appears in threat intel at low confidence, a behavioral anomaly with no corroborating signals — the verdict records this and the confidence level drops accordingly. Low-confidence verdicts go to a security engineer rather than auto-closing.

Step Three — Independent Verification

The triage verdict goes to a second LLM from a different vendor. This model's task is adversarial: find reasons the verdict is wrong. The verifier checks whether cited evidence actually supports the claims made, whether required evidence was gathered, and whether the verdict's confidence level is consistent with what the enrichment data shows.

A verdict the verifier cannot refute passes. A verdict with a gap — a claim that exceeds its evidence, a cited result that does not say what the verdict claims it says, or a required enrichment source that failed — is flagged as disputed. Disputed verdicts go to a security engineer and are not auto-closed or surfaced to the client as resolved.

Using models from different vendors reduces correlated failure modes. If one model has a systematic blind spot for a particular class of alert — a pattern of false confidence on a specific MITRE technique, for example — the verifier from a different vendor is less likely to share that blind spot. This is not a guarantee, but it is materially better than single-model verification.

Step Four — Evidence Sufficiency Check and Routing

Before any alert can be auto-closed, the pipeline checks whether the required enrichment was actually collected. If a lookup failed, timed out, or returned no data, the sufficiency check blocks auto-close regardless of the triage verdict. An alert can only be closed automatically when the evidence the verdict rests on is confirmed to be present and intact.

After the sufficiency check, routing is deterministic. Alerts with benign verdicts, no verifier dispute, and complete evidence are auto-closed with the full trace preserved. Alerts where the verdict is uncertain, the evidence is borderline, or the verifier flagged a concern go to a security engineer's queue with the full enrichment package. High-confidence threat alerts escalate with immediate notification.

Every alert exits the pipeline with its full trace: enrichment results, triage verdict with citations, verifier assessment, sufficiency check outcome, and routing decision. This trace is the audit record for the alert's disposition and the starting point for any further investigation.

What the investigation trace contains:

  • Raw alert fields from Wazuh
  • MITRE ATT&CK technique and tactic from sensor
  • Threat intelligence lookup results per indicator
  • Alert history for the involved entity
  • Identity context for cloud events (MFA, roles, recent logins)
  • LLM triage verdict with per-claim source citations
  • Verifier assessment and any flagged gaps
  • Evidence sufficiency check result
  • Routing decision and timestamp
Common Questions
Can the AI analyst miss something a human analyst would catch?
Yes. A human analyst brings contextual knowledge about an organization that the AI does not have, can recognize emerging attack patterns not yet in enrichment sources, and can make judgment calls about operational risk that the structured verdict cannot fully model. ARIA's design reflects this: anything the AI classifies with low confidence goes to a security engineer. The AI handles the unambiguous cases; humans handle the ambiguous ones.
What does the structured verdict actually contain?
The verdict contains: a severity classification, a disposition recommendation (benign, investigate, escalate), a narrative explanation with each factual claim citing its enrichment source, a confidence level, and a list of any enrichment sources that failed or returned no data. The format is structured enough to be validated automatically but readable enough to be reviewed by a human.
How does ARIA prevent the LLM from making up threat conclusions?
Two ways. First, structurally: the verdict format requires citations, so a claim without a corresponding enrichment result fails validation. The model cannot cite a threat intel result that was not in the enrichment package. Second, through the verifier: the second model checks whether the cited evidence actually supports the claims. A verdict where the citation does not match the source data is flagged as disputed and goes to a security engineer.
What enrichment sources does the AI analyst consult?
MITRE ATT&CK technique from the Wazuh sensor, threat intelligence from OTX and CISA KEV (and other configured feeds), alert history for the entity from the alert database, and — for cloud events — identity context including MFA status, assigned roles, and recent login history. The specific sources active for a given alert depend on the alert type and the integrations configured for that client.
Can I see the reasoning behind a verdict?
Yes. Every alert carries a full trace that includes the enrichment data, the triage verdict with its citations, the verifier's assessment, and the routing decision. The trace is accessible from the alert detail view in the SOC dashboard.

See ARIA in Practice

Book a free assessment to see how the pipeline handles your actual alert environment.

Book Free Assessment
No contract. No setup fee. Cancel anytime.

See Plans

Starting at $799/month. No long-term contracts required.

View Pricing