Run DataKnobs in your own cloud, data platform, and AI environment: or use a DataKnobs-managed environment with prepaid provider budgets and explicit spending controls. The core DataKnobs capabilities remain the same; ownership and operating responsibility change.
Two operating models
Both models can deliver the same governed intelligence. The difference is who owns the cloud and model contracts, who operates the environment, and how provider consumption is billed.
Deploy DataKnobs inside the customer's approved cloud, data platform, model environment, IAM, and security boundary.
DataKnobs hosts and operates the environment while provider consumption stays inside an approved prepaid budget and operating policy.
Decision framework
Customer-owned infrastructure is usually the cleanest fit when workloads contain regulated or highly confidential data, require strict residency, or must remain inside an existing enterprise security boundary.
Deep integration with enterprise warehouses, internal applications, private networking, IAM, and existing cloud operations generally favors customer-owned deployment.
Managed infrastructure can reduce setup and day-to-day operations for focused workloads: provided the security, data-handling, and consumption profile can be explicitly governed.
Commercial model
The two deployment models share the same core platform economics, but managed infrastructure adds a fixed operations layer and a prepaid provider-consumption budget.
One-time discovery, integration, configuration, evaluation, security work, testing, and production launch.
Annual DataKnobs / EKIP subscription based primarily on production Data Products and governed intelligence.
Customer-owned: operated by the customer. Managed: a separate fixed DataKnobs operations fee based on topology, support, and SLA.
Customer-owned: paid directly to providers. Managed: prepaid, metered, and governed against an approved consumption budget.
Side-by-side
| Consideration | Customer-owned infrastructure | DataKnobs-managed infrastructure |
|---|---|---|
| Cloud / LLM contracts | Customer-owned | Operated by DataKnobs within the agreed commercial model |
| Provider consumption | Customer pays providers directly | Prepaid usage budget with transparent metering |
| Environment operations | Customer or its managed provider | DataKnobs |
| Security boundary | Native customer boundary | Defined by scoped managed architecture and agreement |
| Integration depth | Best for complex enterprise integration | Best for standard or intentionally isolated integrations |
| Time to initial environment | Depends on enterprise onboarding and security review | Typically faster for qualified workloads |
| Consumption risk | Customer controls provider budgets directly | Controlled through prepaid budgets, alerts, gates, and stop/degrade policies |
| Typical default | Enterprise / regulated / deeply integrated | Focused, qualified, or operations-light workloads |
Deployment recommendations depend on the specific workload, data classification, contractual requirements, provider architecture, and security controls. The table is guidance, not a blanket eligibility rule.
Portability can be designed into the solution so pilots, public-data products, or focused workloads can start in a managed environment and move later as scale, sensitivity, or integration depth increases.
FAQ
Yes. Both can include governed Data Products, Knobs, lineage, evaluation, monitoring, experimentation, optimization, and EKIP capabilities. The main difference is who owns and operates the infrastructure and provider contracts.
No. It is the default when an organization needs its own security boundary, existing cloud contracts, deep internal integration, or direct control of provider consumption, regardless of company size.
Potentially, but only when the required security, privacy, contractual, residency, and compliance controls are explicitly scoped and approved. Customer-owned infrastructure remains the simpler default for many regulated or deeply integrated enterprise workloads.
Yes. Portability can be designed into the architecture. Migration work, provider contracts, data movement, and security onboarding are scoped separately.
No. Published figures are planning ranges and starting prices. Final commercial terms depend on Data Product scope, integrations, environments, support, security, compliance, and provider-consumption requirements.
DataKnobs can classify the workload, identify security and integration constraints, estimate provider consumption, and recommend the right deployment and pricing path.
Request a deployment assessment