Meet us at #FIRSTCON26 in Denver — June 14–19.Book a booth meeting →

    Arcanna

    ARCANNA DOES |

    Faster, better decisions. Always under human control.
    Automation executes. Decision authority stays governed.

    Analyst-taught, supervised models that deliver consistent, explainable SOC decisions.

    Decision Hero

    DECISION LAYER

    Analyst-taught decision models turn noisy alerts into fast, consistent outcomes by enforcing decisions with guardrails, explainability, and human control.

    Pipelines

    Pipelines

    Structured decision pipelines that move events from ingestion to action with human gating and full control.

    Models

    Models

    Analyst-taught, supervised models that apply consistent decision patterns with full explainability at scale.

    Flows

    Flows

    Execution workflows that apply decisions across tools, from enrichment to action, within defined boundaries.

    Graph

    Graph

    A living decision graph that links signals, decisions, and outcomes for traceability, context, and audit.

    Agents & playbooks execute.
    Control stays human.

    Speed without control breaks trust.

    Arcanna inserts humans exactly where decisions matter:

    90–100% confidence → autonomous execution

    70–89% confidence → human validation

    below 70% → recommendation only

    Autonomy where it's earned.
    Oversight where it's required.
    Arcanna Control Diagram
    Arcanna Grounding Interface

    Agents hallucinate. Arcanna does grounding.

    Agents don't decide in isolation.

    Arcanna grounds every action in people decisions (real time) and organizational ontology:

    historical decisions
    organizational context
    validated human feedback

    Decisions are learned, versioned, and auditable - not improvised.

    Agents act. Arcanna decides.

    Real-world integrations where Arcanna governs decisions before action execution.

    Splunk SIEM
    Splunk SIEM
    Apply analyst-taught decisions before action.
    View integration use case →
    CrowdStrike Falcon
    CrowdStrike Falcon
    Add decision context before endpoint response.
    View integration use case →
    IBM QRadar
    IBM QRadar
    Apply uniform decision logic to offenses.
    View integration use case →
    Swimlane SOAR
    Swimlane SOAR
    Gate workflows with analyst-taught decisions.
    View integration use case →
    Arcanna.ai decides what should happen and how far automation should go. Agents and playbooks execute within those boundaries.

    WHAT ARE DECISION MODELS?

    Models trained to make decisions the way your analysts do.

    Decision models learn from analyst decisions, then apply those decision criteria consistently at machine speed — with human control when confidence is low.

    Learn how Decision Models work →
    01

    Inputs

    Signals and context your analysts trust.

    ↓
    02

    Decision criteria

    Thresholds and rules learned from analyst decisions and applied consistently.

    ↓
    03

    Outcomes

    Predictable decisions, with human validation when confidence is low.

    Agents forget.
    Arcanna
    remembers.

    Every decision becomes memory.

    Arcanna builds a living Decision Graph that captures:

    what happened
    why it was decided
    who validated it
    confidence

    Institutional decisions - finally encoded.

    Decision Models. Built to scale.

    Scalability and Deployment of Arcanna Decision Models

    Deploy anywhere, scale effortlessly:

    CPU-based - no GPU required
    On-prem/private cloud (offline)
    VM and Kubernetes deployments
    AWS and GCP Marketplace availability
    High Availability deployments
    API-first architecture
    Enterprise-grade infrastructure. Your choice of deployment.
    Arcanna Decision Models Dashboard

    How the Decision Layer is introduced safely

    The Decision Layer is designed to be introduced gradually, with zero workflow disruption and full human oversight at every stage.

    1

    CONNECT (~30 MIN)

    Read-only. No workflow changes.

    2

    HITL LEARNING (7-10 DAYS)

    Review decisions. Set autonomy boundaries.

    3

    ENABLE SCOPED AUTONOMY

    Only where confidence is high.

    4

    OPERATIONALIZE

    Monitor KPIs. Adjust over time.

    FAQs ABOUT DECISION LAYER

    What is the
    Decision
    Layer?
    Can agents make
    trustworthy
    decisions?
    How does
    explainability
    work?
    How do I
    audit
    AI decisions?
    Can I
    roll back
    bad decisions?
    Does it require
    GPUs
    to run?
    I already have
    SOAR or n8n.
    Why do I need
    Decision Models?