Security & Data Residency

Your voice AI, deployed where your policy requires.

Architected for self-hosting from day one: strict tenant isolation, conversation and tool guardrails, and flexible deployment from managed cloud to on-premise GPUs.

Three deployment architectures

01

Muwakil Managed Cloud

Launch in minutes with encrypted transport and per-tenant workspace isolation.

  • Zero infrastructure setup
  • Automated backups
  • Fastest path to pilot
02

Private Cloud (Your VPC)

Deploy the runtime inside your dedicated cloud network connected privately to internal CRMs.

  • Private networking & IAM
  • Isolated storage & logs
  • Built for regulated teams
03

Self-Hosted On-Premise

Run the complete speech and agent stack on your own GPU infrastructure.

  • 100% data residency
  • Zero external audio calls
  • Full infrastructure control

Built-in platform controls

Tenant & key isolation

Each workspace isolates its own agents, knowledge bases, custom voices, conversation logs, and hashed Bearer API keys.

Input & output guardrails

Screens for prompt injection, masks sensitive PII, and restricts policy/pricing replies strictly to your uploaded KB.

Verified voice-cloning consent

Custom voices require explicit speaker authorization and stay private to your tenant—never exposed in the public voice library.

Full data ownership

Your organization owns every call recording, transcript, and structured outcome, with API access to inspect or delete.

Explicit confirmation for actions

Sensitive tool actions (refunds, cancellations, order edits) require explicit caller confirmation before execution.

Self-hosted or private VPC deployment

Deploy the full voice and agent runtime on your own GPUs or private network with zero external audio transit.