Pensero supports both managed SaaS and on-premise deployments. SaaS is the standard option. Because on-premise environments require materially more deployment and maintenance work, that option is available for customers above an appropriate deployment size and is assessed during technical scoping.

In the SaaS deployment, Pensero operates the application and its AI infrastructure. Pensero's models run on Amazon Bedrock inside Pensero's AWS account. For details about what Pensero processes, sends to AI, and stores, see What data Pensero processes, sends to AI, and stores.

In an on-premise deployment, the customer operates the compatible model endpoints and the infrastructure behind them. Pensero does not deploy or manage a GPU cluster.

The environment must expose the same supported models used by Pensero's agents. Before approving the integration, Pensero runs a quality evaluation to confirm that agent behavior remains consistent with the expected results. The final application topology, security controls, and operational responsibilities are agreed with the customer during technical scoping.