# Causal Security Intelligence for SIEM and SOC **Source:** [TRA-CE](https://tra-ce.ai) --- # Know what happened. Prove why. Your SIEM tells you what. It never tells you why. TRA-CE sits beside it and builds the causal chain — every link evidence-graded, every finding traced to source, every conclusion an analyst can defend in front of an auditor. ## How it works 01 INGEST ### Deploy the collector. A lightweight Docker container inside your network, encrypted end-to-end via Avon — our post-quantum transport layer using ML-KEM-768 + X25519 hybrid key exchange with 30-second session rotation. It connects to your SIEM, EDR, identity provider, or cloud platform. No agents on endpoints. 02 CHAIN ### Build causal chains. Events get linked into directed causal graphs using four inference heuristics: explicit system evidence, artifact correlation, MITRE technique progression, and temporal proximity. Each link carries a confidence grade — PROVABLE, MIXED, or INFERRED. Nothing is asserted by default; everything is graded. 03 ACT ### Investigate with evidence. AI investigation agents run a three-stage pipeline: gather evidence, build the WHY stack, map interventions. The output is a structured argument an analyst can read in minutes — what failed, what should have caught it, what to fix, who owns it. Analysts decide. TRA-CE provides the proof. ## What's under the _hood_ Core Engine ### Causal chain building. Four inference heuristics construct directed evidence graphs from event streams: explicit system evidence, artifact correlation, MITRE technique progression, and temporal proximity. Each edge is graded independently. Detection ### Campaign scoring. Related events get grouped into campaigns scored by aggregate evidence weight. The output is not a list of alerts. It is a list of stories. Evidence ### Grade every link. PROVABLE, MIXED, or INFERRED — every edge carries a confidence grade. The chain is only as strong as its weakest link, and the dashboard never lies about which link that is. Investigation ### AI agents do the slog. Investigation agents run a three-stage pipeline: gather evidence, build the WHY stack, map interventions. The agents do the work analysts would do; the analysts review the work the agents did. Compliance ### Audit-ready by default. Every chain is a record. Every grade is a record. Every conclusion is traceable to source events. When the auditors arrive, they get the same chain the analysts built. Transport ### Post-quantum from day one. Avon, the transport layer underneath TRA-CE, uses ML-KEM-768 + X25519 hybrid key exchange with 30-second session rotation. CNSA 2.0-aligned. FIPS 203/204 cryptography. Integration ### Works with what you have. Splunk, Datadog, your existing SIEM and observability stack. Active Directory, Okta, your identity provider. AWS, Azure, GCP. One collector. No rip-and-replace. ## See your event stream as a graph. Point TRA-CE at a slice of your existing telemetry. Walk through what it surfaces. The fastest way to evaluate the product is to evaluate the product. Software for the things that matter.