Your intelligence layer
Your application is not your model—and it is not your cloud.
Keep the system definition stable while choosing the intelligence and environment that fit each workload.
Model architecture
Three independent layers.
Application
Solutions, skills, tools, policy, UI
Intelligence
Hosted APIs, private endpoints, local inference
Infrastructure
Managed cloud, private cloud, on-premises
Connect OpenAI, Anthropic, Ollama, SGLang, and compatible model endpoints—each layer changes without disturbing the others.
Why it matters
Model freedom, in practice.
Choose capability by workload
Keep sensitive workloads private
Optimize latency and cost
Move as the model market changes
Avoid application rewrites
Deployment modes
Run it where it fits.
A-Team Cloud
Fastest start and managed operations.
Self-hosted A-Team
A platform instance in customer-controlled infrastructure.
Hybrid
VerifyingHosted and private model endpoints coexist.
One-click deployment
The real self-host flow.
Reserved for a capture of the actual flow, not an abstract cloud graphic.
01Environment selection
02Configuration inputs
03Deployment progress
04Health result
05Connect model endpoint
06First solution test