Your cloud
The runtime and agent run in your AWS, Azure or GCP account.
- Your team sets network and identity controls
- Connect the runtime to your existing logs and cost controls
- Choose a hosted model or one you serve privately
Deployment & data control
Your cloud or ours. Your call.
We build the agents. You choose whether the platform runs in your cloud, on your own infrastructure or in a Koru-managed environment. You also choose which models they call.
Discuss your deployment requirementsThree ways to run
Where it runs
Your infrastructure
Run on your hardware or in your data centre, including air-gapped environments.
- Keep the runtime and indexes inside your network
- Use an open-weight model locally when no egress is allowed
- Plan capacity and operations with your infrastructure team
Managed by Koru
We host and operate the deployment when you want us to handle the infrastructure.
- Koru operates the agreed environment
- We document access, retention and model data flows
- Review a future move to your environment when useful
The decision behind the decision
What leaves your network?
A customer-cloud deployment keeps the runtime and its records in your account. If that agent calls a hosted model, the selected request data goes to that provider. For a workflow with no permitted external egress, run an open-weight model inside your environment.
Review security and data handling| Choice | Where agent records live | Model path |
|---|---|---|
| Your cloud | Your account | Your selected provider or private model |
| Your infrastructure | Your network | Local model for no-egress workflows |
| Managed by Koru | Koru-managed environment | Agreed provider or private model |
Ownership
You own the agents we build for you.
Koru Runtime is the platform that runs them. We define its deployment, access and ongoing terms in the engagement, so your team knows what it owns and what Koru operates.
A useful first conversation
Where does the work get stuck?
Bring the workflow people actually follow, including the workarounds. We will trace where it gets stuck. Then we will agree on the approvals and one result to measure after launch.
Map a workflow with our engineers
Talk directly with the engineers who build the agent.