Deployment
Where the data sits is the decision that matters most
When feature lists converge, what actually decides feasibility is where the data lives. EgentWrX runs on cloud, on-premises or hybrid infrastructure, and this page sets out how the three differ and which conditions suit each.
Three architectures
IT resources, data sensitivity, compliance and use cases differ at every company. No single architecture is best for everyone.
We host it; nothing to build or maintain
For companies that want speed, or are running an early trial with limited resources. No local hardware, live on activation, with operations and updates handled by us.
- Usable the moment the account opens
- No hardware purchase, lowest upfront cost
- No dependence on an internal IT team
Data and models stay inside the company
For companies with strict control requirements or in regulated industries. Storage and access are entirely yours. Setup costs more, but the architecture can be shaped around your environment.
- Data and compute both on site
- Full ownership of access and governance
- Architecture adapts to your existing environment
Data on site, inference in the cloud
The most sensitive data stays on premises or in a private cloud while model inference runs in the cloud, giving you frontier compute. Companies with strict security needs that still want the latest models usually choose this.
- Sensitive data never leaves, top-tier models still available
- Switch between large language models at will
- Balances flexibility, security and cost
How to choose
This table compares the three architectures on the conditions that usually decide it. EgentWrX EDGE, the edge computing module, layers onto any of them; it is not a fourth architecture.
| Hybrid | Cloud | On-premises | |
|---|---|---|---|
| Character | Data on site, compute in cloud | Data and compute in cloud | Data and compute on site |
| Time to launch | Fast | Fast | Longer |
| Data residency | Private cloud or on site | Cloud | On site |
| Upfront cost | Moderate | Lower | Higher |
| Model choice | Wide | Wide | Local models only |
| Model capability | Frontier | Frontier | Moderate |
| EgentWrX EDGE | Available | Available | Available |
Hybrid
- Character
- Data on site, compute in cloud
- Time to launch
- Fast
- Data residency
- Private cloud or on site
- Upfront cost
- Moderate
- Model choice
- Wide
- Model capability
- Frontier
- EgentWrX EDGE
- Available
Cloud
- Character
- Data and compute in cloud
- Time to launch
- Fast
- Data residency
- Cloud
- Upfront cost
- Lower
- Model choice
- Wide
- Model capability
- Frontier
- EgentWrX EDGE
- Available
On-premises
- Character
- Data and compute on site
- Time to launch
- Longer
- Data residency
- On site
- Upfront cost
- Higher
- Model choice
- Local models only
- Model capability
- Moderate
- EgentWrX EDGE
- Available
For companies balancing flexibility, security and cost, we usually suggest starting hybrid.
Where to go next
Want every employee to have their own AI teammate?
Places are limited. Our team will be in touch after you register.