Deployment
Models
Intelligence that adapts to your environment — not the other way around.
Genovation supports multiple deployment models without changing the intelligence layer, governance guarantees, or security posture. All powered by Mentis OS.
4
Models
100%
Parity
0
External APIs
Powered by Mentis OS
Unified control plane across all deployment models
Design Philosophy
Deployment as a
First-Class Constraint
Most AI platforms optimize for cloud-first execution and retrofit enterprise options later. Genovation takes the opposite approach.
Typical Approach
Genovation Approach
Supported Models
Four Deployment Models
Choose the model that fits your infrastructure. All deliver identical intelligence, governance, and security.
On-Premise
Data Sovereignty
Deployed entirely within customer infrastructure. No outbound data movement. Air-gap compatible.
Customer Infra
Zero Outbound
Hardware Control
Air-Gap Ready
Private Cloud
Controlled Environment
Deployed within private VPCs or enterprise networks. No shared inference. Controlled ingress/egress.
Private VPCs
Dedicated
Tenant Isolation
Traffic Control
Hybrid
Balanced Flexibility
Core intelligence on-premise, select integrations across zones. Unified governance via Mentis OS.
Local Core
Selective Sync
Cross-Zone Gov.
Central Oversight
Air-Gapped
Maximum Isolation
No external connectivity. Offline inference and agent execution. Secure artifact delivery.
Zero Network
Offline Inference
Secure Updates
Full Audit
Full Architecture
Deployment Architecture Stack
A complete view of all layers — from infrastructure through intelligence to integration — unified by Mentis OS.
Application
APIs & Integration
Governance
Policy & Audit
Intelligence
SLMs & Agents
Mentis OS
Control Plane
Infrastructure
4 Deploy Models
Web Console
Admin Dashboard
REST API
External Interface
SDK
Python / JS / Go
CLI Tools
Command Line
Webhooks
Event Callbacks
Stream API
Real-time Output
Observability
Metrics / Logs
Policy Engine
Rules / Constraints
Access Control
RBAC / Permissions
Audit Logger
Immutable Records
Explainability
Decision Traces
Compliance
Standards / Reports
SLM Models
Reasoning / Planning
Domain-Optimized
Agent Runtime
Task Execution
Multi-Step Workflows
Tool Framework
Function Calling
Sandboxed Execution
Context Manager
Memory / State
RAG Integration
Orchestrator
Multi-Agent Coord.
Pipeline Control
Scheduler
Task Queue
Resource Mgr
Allocation
Config Store
Versioned State
Health
Liveness / Ready
Secret Mgr
Credentials
Update Ctrl
Rollout / Rollback
Telemetry
Collect
Container
Docker / Podman
K8s / Nomad
Orchestration
Storage
Local / S3 / NFS
Networking
Service Mesh
Load Balancer
Traffic Mgmt
Registry
Images / Artifacts
DNS
Discovery
GPU Compute
CPU Compute
Memory
Block Storage
Network I/O
HSM / TPM
Edge Nodes
Deploys identically across: On-Premise · Private Cloud · Hybrid · Air-Gapped
Consistency
Identical Across All Models
No "lite" versions. Every deployment model delivers the same guarantees.
Intelligence
Identical
Governance
Identical
Security
Identical
Audit
Identical
| Capability | On-Prem | Private | Hybrid | Air-Gap |
|---|---|---|---|---|
| Model Inference | ||||
| Agent Orchestration | ||||
| Policy Enforcement | ||||
| Explainability | ||||
| Audit Trail | ||||
| Mentis OS Control |
Enterprise-Friendly Infrastructure
SLM strategy avoids hyperscaler dependency.
Commodity GPUs
Standard enterprise hardware
CPU-Only Options
For specific workloads
Scalable Multi-Node
Grow with requirements
No Vendor Lock-In
Full ownership. No proprietary dependencies.
Not Required
You Retain
Long-term strategic flexibility guaranteed
The Principle
"Intelligence should adapt to the enterprise — not force the enterprise to adapt to intelligence."
This defines Genovation's deployment philosophy.