01Built for the conversations that have to work
For suppliers, factory floors, doctors, customers and partners on the other side of the world — where the network is whatever it wants to be, and the meeting can't be rescheduled. SpeechKey gives international teams a single, calm translation layer they can trust.
02BYOK at scale
- Centralised key management across OpenAI, Qwen, xAI Grok and regional providers.
- Per-team or per-region key allocation.
- Automatic key rotation with audit trail.
- Spend visibility against each provider, broken down by team, region and use case.
- Token costs stay between your organisation and your providers — we don't mark them up.
03Custom & private AI providers
- Bring your own provider endpoint, including private deployments of OpenAI on Azure, Vertex AI on Google Cloud, or self-hosted open models behind a compatible API.
- Custom voice profiles for specific languages or speakers, where supported by your provider.
- Per-tenant model allow-list and provider routing rules.
04Deployment options
- SaaS
- EU-default. US, APAC and CN regions available.
- Dedicated tenant
- Single-tenant database and infrastructure within your selected region.
- Private network
- VPC peering and PrivateLink-style connectivity to your AI providers.
- On-premise
- For regulated industries — pilot programme available. Contact us.
05Security & governance
- TLS 1.3 in transit, AES-256 at rest, customer-managed keys (CMK) available.
- Hardware-key MFA enforced for admin roles.
- Tamper-evident audit log streaming to your SIEM.
- Quarterly access reviews; annual third-party penetration tests.
- GDPR-aligned; SOC 2 Type II audit programme in progress.
See the dedicated Security page.
06Procurement readiness
- Master Services Agreement (MSA) and Data Processing Agreement (DPA) templates available.
- Standard Contractual Clauses (SCCs) for international transfers.
- Pre-completed SIG and CAIQ-Lite questionnaires.
- Vendor security review turnaround: 3 business days.
- Insurance certificates, sub-processor list and incident history available under NDA.
07Scalability
- Horizontally scaled realtime layer with automatic regional load balancing.
- Provider-level rate-limit pooling so one busy team doesn't choke another.
- SLOs: 99.9% availability target for the resilient layer; provider-specific SLOs depend on the underlying provider.
08Admin controls
- SSO (SAML 2.0, OIDC) and SCIM user provisioning.
- Role-based access control: owner, admin, billing, audit-viewer, end user.
- Tenant-wide policy for transcript retention and analytics opt-in.
- Region-pinning per team — e.g. CN team uses iFLYTEK; EU team uses OpenAI EU; false-route attempts are blocked.
- Centralised provisioning of API keys; end users never see raw credentials.
09Data ownership
- You own your transcripts, conversations and configuration — always.
- We do not use your data to train models. Ever.
- One-click export of all tenant data in JSON.
- Legal-hold and tenant-purge actions available to admins.
10Integrations
- SIEM (audit-log streaming) — Splunk, Datadog, Elastic.
- SSO / IdP — Okta, Entra ID, Google Workspace, Auth0.
- Webhook events — session start/end, transcript export, security events.
- Calendar & meeting tools — Zoom, Teams, Google Meet via captured-audio bridge (beta).
11DPA & legal
DPA, SCCs, sub-processor list and security exhibit are available on request from legal@vaitl.ai. Most enterprise customers can close paperwork in under two weeks.
12Contact sales
- Enterprise sales
- enterprise@vaitl.ai
- Trust & compliance
- trust@vaitl.ai
- Security questionnaires
- security@vaitl.ai