Design preview

What we are looking for

The current design-preview fit is a team already using, evaluating, or seriously considering Turbopuffer, with a retrieval workload large enough to need an operating layer but small enough for a tightly supported first deployment.

• Under 3.5 TB of managed source data for the first engagement.
• No customer-managed encryption key requirement right now.
• A concrete read path that needs fetch, cache warming, scans, snapshots, or pipeline state.
• A team willing to share operational feedback on cache behavior, cutover, and pipeline ergonomics.

What you get

• A hosted hev layer gateway pointed at your Turbopuffer workload.
• Private access to the hev-shop source as a reference starter kit — fork and customize the pipeline, workers, Helm chart, and Layer client for your workload.
• Helm-managed deployment operated with you, not a self-serve platform.
• Docs and examples focused on your integration path.
• Direct feedback loop while the product surface is still narrow.