zoé

Architecture

How Zoé fits into your stack.

Four cleanly-separated layers, one autonomous agent. Zoé sits alongside your existing ITSM and identity systems — augmenting, not replacing, the tools you already trust.

01
Layer 01
Experience
Voice
Chat
Biometrics
02
Layer 02
Zoé Core
Reasoning
Orchestration
Guardrails
03
Layer 03
Integrations
Azure AD
Nexthink
Zscaler
ServiceNow
Intune
M365
04
Layer 04
Infrastructure
Knowledge
Model layer
Observability
Data flow
Zoé core
Layer 1

Experience layer

Where users meet Zoé

  • Voice assistant
    Hands-free L1 triage over phone / Teams
  • Chat & web portal
    Conversational self-service across devices
  • Voice biometrics
    Identity via enrolled speech signature
Layer 2

Zoé core

The autonomous brain

  • Reasoning engine
    Intent, classification, confidence gating
  • Runbook orchestrator
    Deterministic multi-step automations
  • Guardrails
    Policy, privacy, safe-action enforcement
Layer 3

Integration layer

The enterprise connectors

  • Nexthink
    Endpoint telemetry & remote actions
  • AD / Azure AD / Okta
    Identity, MFA, lifecycle, SSO
  • Zscaler / ZPA
    VPN & ZTNA tunnel diagnostics
  • ITSM (ServiceNow)
    Tickets, approvals, CMDB sync
  • Microsoft Intune
    Policy, app deployment, patching
  • M365 / Graph
    Mail, Teams, calendar, user data
Layer 4

Infrastructure

Secure, observable, scalable

  • Knowledge & KB store
    RAG over runbooks & ITIL policies
  • Model layer
    On-prem / private-cloud LLMs, speech models
  • Observability
    Audit, SIEM feed, full action trail

A request, end-to-end

From the user’s voice to the resolved ticket.

  1. 1
    Voice / chat in
    User speaks or types the problem.
  2. 2
    Intent & policy
    Zoé classifies, checks entitlements and gates.
  3. 3
    Orchestrate
    Runbook fires across AD, Intune, Nexthink, Zscaler.
  4. 4
    Validate
    Post-checks confirm resolution, health score, etc.
  5. 5
    Close & audit
    Ticket closes, full audit trail to SIEM.