Deployments

Deployments is the pillar that answers procurement's first question: where does our data live, and who can touch the machines? The default answer is the managed cloud — multi-tenant with kernel-enforced isolation, regionally pinned so a project created in ap-south stays in ap-south for compute, storage, and traces alike, and covered by the SLA on the status page. One tier up, dedicated capacity gives regulated customers single-tenant compute and storage in their chosen region with the same API surface, so nothing in your integration changes when compliance upgrades your tier. At the strict end, self-hosted Mynd runs the full platform — runtime, context graph, trust kernel, observability — inside your own VPC or on-premise cluster, delivered as signed containers with an offline-capable license, taking updates on your schedule through a staged channel you control. The promise across all three is API identity: the same code, keys model, and traces everywhere, so a pilot on managed cloud moves to self-hosted by changing a base URL. Operational posture is published rather than implied — deployment options, regions, isolation guarantees, and update cadences are documented, environments are reproducible from declarative config, and the migration paths between tiers are tested product surfaces with their own docs, not professional-services folklore.