Your infrastructure
Review capacity, networking, physical access, backups and the people who maintain the hardware. Decide who can support the application and how updates reach the environment.
Deployment and technical diligence
Define the boundary before calling AI private.
Private AI can run on your infrastructure, in your cloud account or in a managed environment. The deployment choice alone does not settle access, retention or operating risk. Review the complete application and its dependencies.
Request a project reviewTell us what needs to change. We’ll review the context and discuss the next step with you.
Three environments, different responsibilities
Questions a security review should answer
Where does data travel?
Map prompts, source files, retrieved content, outputs, telemetry, logs and backups. Name external services, their purpose and the applicable retention settings.
Who can access it?
Describe identities, roles, privileged access, approval and revocation. Review support access separately from normal application use.
What remains after a request?
Define logging, storage, retention, deletion, backup recovery and whether any data is used to improve a model. Confirm the actual configuration and contractual terms.
Who handles an incident?
Assign monitoring, triage, notification, recovery and change responsibilities. Confirm the agreed service coverage and the evidence retained for investigation.
Ask for evidence tied to the actual environment.
A facility’s attestation is not an attestation for an application or for every service in its data path. Review the entity, covered service, period, exclusions and customer responsibilities for any control report being relied on.
Review the applicable architecture, access arrangements, dependency inventory, operating responsibilities and available control evidence for the proposed scope. Confirm each control against the actual application and operating agreement.
Plan for operating change and handoff.
Models, dependencies and workloads change. Agree who approves updates, tests regressions, reviews cost and handles rollback. For a handoff, identify the configuration, code, runbooks, evaluation assets and licenses your team will receive.
Read the platform ownership field report · See the private AI offer · Inspect delivery decisions
Discuss the work you need to change.
Describe the business problem, the systems involved and the constraints that matter. We’ll review the context and discuss a useful next step.
Request a project reviewPlease keep confidential documents, credentials and personal data out of the initial inquiry.
