Full Data Sovereignty
All data stays within your data center. No call recordings, transcripts, or customer data ever leave your network.
On-Premise AI Voice Agents — Your Data Center, Your Rules
isLucid's Agent Box deploys directly in your data center. Zero cloud dependency, full data sovereignty, SOC 2 + ISO 27001 certified. Live in 24 hours.
Zero Cloud dependency
How It Works
Step 1: Ship Agent Box to your data center
Receive pre-configured hardware or deploy on your existing infrastructure via Docker/Kubernetes.
Step 2: Connect to your phone system
SIP trunk integration with your existing PBX, contact center, or VoIP infrastructure.
Step 3: Train on your scripts
Upload call scripts, FAQs, and knowledge base. AI learns your business locally.
Step 4: Go live in 24 hours
Start handling real calls with full data sovereignty. Scale by adding more nodes.
Benefits
Sub-10ms Latency
Local processing eliminates network round-trips. Voice AI responds faster than cloud solutions.
Regulatory Compliance
Meet the strictest compliance requirements — HIPAA, GDPR, data residency laws — with full on-premise control.
No Internet Dependency
Operates fully offline once deployed. Network outages don't affect your AI voice agents.
Technical Specifications
- Docker/Kubernetes deployment
- ARM64 + x86_64 architecture support
- Minimum 16GB RAM, 4 vCPU per node
- GPU optional (faster inference with NVIDIA CUDA)
- API-compatible with cloud version
- Auto-updates with one-click rollback
- Horizontal scaling — add nodes as needed
- Monitoring dashboard with health checks
On-Premise vs Cloud Deployment
| Metric | On-Premise (Agent Box) | Cloud |
|---|---|---|
| Data location | Your data center | Cloud provider |
| Latency | <10ms | 50-200ms |
| Compliance control | Full control | Shared responsibility |
| Internet required | No | Yes |
| Cost model | Fixed infrastructure | Per-minute usage |
| Scaling | Add nodes | Automatic |
Frequently Asked Questions
What is on-premise AI deployment?
On-premise AI deployment means running the AI voice agent software entirely within your own data center or infrastructure. With isLucid's Agent Box, no data — call recordings, transcripts, or customer information — ever leaves your network.
How does the Agent Box work?
The Agent Box is isLucid's on-premise deployment solution. It can run on pre-configured hardware we ship to you, or deploy on your existing infrastructure via Docker or Kubernetes. It includes the full AI voice agent stack: speech recognition, natural language understanding, text-to-speech, and call management.
What hardware is required?
Minimum requirements are 16GB RAM and 4 vCPU per node. GPU (NVIDIA CUDA-compatible) is optional but provides faster inference. The system runs on both x86_64 and ARM64 architectures. Most modern server hardware meets these requirements.
Can on-premise scale like cloud?
Yes. The Agent Box supports horizontal scaling — add more nodes to handle more concurrent calls. While cloud scales automatically, on-premise scaling requires adding infrastructure, but the process is straightforward with Kubernetes orchestration.
Is the on-premise version feature-equivalent to cloud?
Yes. The Agent Box runs the exact same AI models and features as our cloud version. API compatibility means you can develop against the cloud API and deploy on-premise without code changes. Updates are delivered the same way.