Home › Platform › Product tour

From the packet on the wire to the number on the board

Two sensors, one console, eight views. This is what an analyst, an architect, and a board member each see — the same telemetry, read at the altitude each of them works at. Every screen here is the live product, captured as it runs.

Prefer to watch it? Three unedited demos live in Resources →

One platform · two sensors

Two sensors, one decision loop — one at the browser plane, one at the system plane. Neither sinks an agent into your endpoints.

Two sensors run on one endpoint and feed one decision loop, the CognitionAI Engine. The Browser Sensor runs in the browser session, in-page, reading prompts, GenAI sessions and tool calls across Chrome, Chromium, Edge and Safari, and acts at the point of interaction. The System Sensor runs in userspace on native OS APIs, reading processes, agents and AI-created artifacts across Windows, macOS and Linux, and acts at the point of execution. Neither adds kernel surface: no driver, no kernel extension, no EDR hook, no Ring 0 component, and neither requires root or administrator privilege.
One decision loopCognitionAI Engine
Browser planeBrowser SensorSystem planeSystem Sensor
Where it runsIn the browser session, in-pageIn userspace, on native OS APIs
What it readsPrompts · GenAI sessions · tool callsProcesses · agents · AI-created artifacts
Kernel surfaceNone — the extension runs in-page, not on the hostNone — no driver, kernel extension, EDR hook or Ring 0
Privilege requiredNone beyond the extensionNo root, no administrator
PlatformsChrome · Chromium · Edge · SafariWindows · macOS · Linux
Where it actsAt the point of interactionAt the point of execution
Two planes. One decision loop. No kernel agent on either.
Multi-persona views · one console

From the packet on the wire to the number on the board — every layer of the user edge, in one place.

8 views · scroll or select
0102030405060708
  • 01 · See
    Console enclave graph with Claude, ChatGPT, Gemini, Grok, Copilot and Perplexity as linked nodes among devices, people and external destinations, above a counter band reading twenty devices, three people and six AI services.

    Live AI Traffic Topology

    AI services, devices, people and external destinations drawn as one continuously updating map — sanctioned AI and shadow AI side by side.
  • 02 · Attribute
    Network Sensor table of process-attributed egress flows with counters reading 188,116 flows, 94 attributed, 10 AI flows and 6 evaded, and one row flagged as evaded at high severity.

    Flow-to-Process Attribution

    Every outbound connection tied to the local process that opened it — and the ones that bypassed the proxy counted out on their own.
  • 03 · Comprehend
    Connection details panel showing a captured AI prompt and response alongside anomaly, sentiment and hostility scores.

    Intent Inside the Interaction

    The captured exchange, scored for anomaly, sentiment, and hostility — read at the semantic layer rather than matched against a signature.
  • 04 · Verdict
    Anomaly Detected panel scoring a flow high at 65 percent salience, with a neuromorphic breakdown weighting destination novelty, data volume, timing and off-proxy egress, above the source, process, host and detection time.

    The Drivers Behind the Score

    One salience score per flow, broken into the weighted signals that produced it — destination novelty, data volume, timing, and off-proxy egress.
  • 05 · Evasion
    High severity prevented-threat panel reading evaded monitoring for off-proxy egress, with a cognition network verdict, ATT&CK command-and-control and exfiltration tags, potential impact for user, device and organisation, and recommendations.

    Caught Going Off-Proxy

    A process that routed around monitoring — with the behavioural reasoning, the ATT&CK mapping, and the impact stated in one panel.
  • 06 · Provenance
    AI-Created Artifacts table listing files an AI wrote, cloned or dropped — each tagged with the AI tool that produced it, its action, type and size — above network sensor flow and prevented threat panels.

    Files Traced to the AI That Made Them

    Every AI-created artifact recorded at birth — so if it ever executes, the egress is already tied to the session that wrote it.
  • 07 · Enforce
    AI usage view listing detected AI vendors as sanctioned or shadow, with a block action scoped to autonomous activity only.

    Shadow AI, Sanctioned or Blocked

    Vendor-by-vendor usage with sanction and revoke — and enforcement scoped to autonomous activity, leaving human workflows untouched.
  • 08 · Board
    Global overview command centre showing sensor fleet status, a threat prevention engine count of 144,056, sensor geography by region, and threat classification bars.

    The Number on the Board

    Fleet health, prevention posture, and global threat distribution — the same telemetry, in the language of the board.

Bring it to your own environment.

Your traffic, your AI tools, and the same console you just toured — walked through live with an engineer.

Request a Demo