Agent-native telemetry
Structured events for tool calls, approvals, retries, parameters, and outcomes—linked to identity and session context.
Cortex Protect
XDR for AI agents,
not just endpoints.
Your perimeter moved. Cortex Protect is extended detection and response for autonomous systems—so toolchains, approvals, and outcomes show up in the same investigation as hosts and identities.
Structured telemetry Tool calls with identity context
One timeline Agents + endpoints correlated
Guardrailed response Automations with approvals
Live correlation preview
Design goals
- Evidence you can explain to legal and leadership
- Automation that respects approvals and blast radius
- Plays nicely with the SIEM you already fund
Illustrative UI · Not live customer data
Why now
Endpoints were never the whole story.
Agents don’t “live” on a laptop—they chain tools across SaaS, data stores, and infra. If your detections stop where the host stops, you’re blind to the fastest-moving attack surface in your org.
- Chain visibility
- See multi-step tool usage as one narrative.
- Shared context
- Tie prompts, approvals, and outcomes to principals.
- Faster MTTR
- Contain autonomous misuse without waiting for “patient zero”.
- Audit posture
- Prove what was allowed, denied, and overridden.
Platform
Everything you need to treat agents like infrastructure.
Cortex Protect is opinionated about schema and timelines—so detections are explainable and automation is enforceable.
Cross-signal correlation
Join agent actions with endpoint, identity, and cloud signals into one timeline your analysts can actually defend in court.
Guardrailed response
Automations that revoke tokens, isolate hosts, and block tools—with approvals, audit trails, and rollback paths.
Evidence-first investigation
Detections surface rationale and citations: what changed, who approved it, and which policy was in effect.
Workflow
From signal to containment—without losing the plot.
Analysts get a story. Automations get guardrails. Leaders get receipts.
- 01Ingest
Stream agent + infra signals with schema you can version.
- 02Correlate
Map toolchains to users, devices, and change windows.
- 03Decide
Triage with ranked context—not a wall of raw logs.
- 04Contain
Execute playbooks with human gates where it matters.
Pricing
Start with a pilot. Scale with conviction.
Early teams get hands-on support while we harden connectors and playbooks for your environment.
Pilot
For teams shipping agents to production with clear security goals.
- Core ingestion + timeline
- Detection packs (starter)
- Weekly working sessions
Enterprise
For regulated environments and multi-team SOCs that need custom policies and integrations.
- Advanced automation + approvals
- Dedicated support & reviews
- Security reviews & roadmap input
Questions, answered plainly.
Is this replacing our XDR or SIEM?
Cortex Protect is the agent layer. It amplifies what you already run by making autonomous behavior first-class in detection and response—without forcing a rip-and-replace.
What do you mean by “AI agents” here?
LLM copilots, agent frameworks, and autonomous workflows that call tools against your data and systems. If it can act, it should be in your security model.
How do you handle false positives?
Policies, approval trails, and multi-signal correlation reduce noise. Every alert is designed to answer “why now?” with evidence, not just a score.
Are you ready for production?
We’re onboarding design partners. If you’re running agents in production, we should talk about your threat model and integrations.
Contact
Tell us what you’re securing.
We’re onboarding a small number of partners. Share your stack and constraints—we’ll follow up with next steps.