Investigations ship enabled for new tenants. Edg3 assembles bounded evidence from the lakehouse and asks the configured reasoning provider for a structured verdict. Analysts see the cited evidence, ATT&CK context, confidence and gate result, and every investigation is labelled with the signal that triggered it. Analysts then record their own disposition, which is stored separately from the underlying events.
Edg3 does not make product authority depend on a model name. The investigation service builds a structured evidence pack, applies a versioned prompt contract, validates the response schema and subjects the result to deterministic grounding and approval rules.
Detections, reusable query results and retrieved context are presented as bounded evidence rather than unrestricted access to the tenant.
Investigations use a model server that Edg3 operates in its own AWS environment and shares across tenants. No outside AI provider is used.
The service records the model used and preserves one verdict schema across model versions.
Grounding checks, rate limits, confidence policy and human approval determine what can progress. The model does not gain direct response authority.
A proposed verdict is useful only when the analyst can inspect its basis. The investigation record includes the available evidence, ATT&CK context, confidence, provider identity and deterministic gate result.
The displayed fields depend on the evidence available for that case. Analysts can use the retained context to reproduce pivots and compare the proposal with the underlying events.
Analyst dispositions (true positive, false positive or benign) are recorded against the incident. They provide review evidence and can inform a separately governed model evaluation process; they do not automatically retrain a production model.
Detection engine fires on a rule or anomaly. Supported endpoint verdicts and cloud-side analytics detections enter one tenant-scoped investigation path. The Edg3 cloud pulls cross-source context (endpoint, network, cloud, identity) from the lakehouse.
The provider returns a structured verdict with technique context, evidence references and confidence. Invalid or ungrounded responses fail the deterministic checks.
The analyst reviews the cited evidence and marks the incident true positive, false positive or benign.
The analyst’s latest disposition, with who set it and when, is stored for review and reporting. The AI’s own outputs go to an append-only log of recent investigations for controlled evaluation.
Prompt, model or policy changes can be tested against retained cases before a new version is selected for production.
Investigation context is sent only to the inference endpoint selected for the tenant. Every AI feature runs on an open model that Edg3 hosts in its own AWS environment and shares across tenants, so investigation data stays inside AWS. Investigations use no outside AI provider. Automated investigation calls pass through an observability layer that records latency, usage and outcome. Credentials and other secret-shaped values are redacted from that retained telemetry by default; user names, host names, IP addresses and command lines are kept as observed. Recursion controls stop the platform investigating its own investigation traffic. If the inference path is unavailable, detection and alerting continue unchanged; alerts that arrive during the outage go to analysts without an AI verdict and are not re-investigated automatically. A fully self-hosted, no-egress deployment remains on the roadmap. No-egress · Q4 2026
The investigation service gathers repeatable query results and cited context before the analyst opens the case. This shifts effort from manual context assembly toward validation, judgement and response. Results still depend on telemetry quality, query latency and the configured inference provider.
For MSSPs, the same structure can standardise how investigations are assembled across the tenants an analyst is authorised to review, while keeping the underlying records scoped.