When Private AI fits
- You are a developer or platform team that needs a redaction API to embed in your own application or data pipeline.
- You need to redact PII/PHI/PCI across many languages and file types programmatically, at scale.
- You can run the service in your own environment (on-premise or VPC) and want full control over deployment.
When Locke fits
- You want a finished end-user control that acts when a person pastes into ChatGPT, Claude, or another AI tool, not an API to build with.
- You want detection and masking to run on the individual's device with nothing to integrate or operate.
- You want a per-user Early Access desktop app (Locke), free for one person, rather than usage-based API infrastructure.
Side-by-side
| Dimension | Locke | Private AI |
|---|---|---|
| Product shape | End-user privacy layer at the point of AI use. | Developer API / infrastructure for redaction pipelines. |
| Who deploys it | Individuals and teams install and run it themselves. | Engineering teams integrate it into apps or workflows. |
| Where processing runs | On the user's device, before the prompt is sent. | In the cloud or in your own on-premise / VPC deployment. |
| Entry point | Locke Free costs nothing (Early Access); Teams starts at $14.99 per seat a month. | API service. Contact vendor for current pricing and terms. |
| Best for | Stopping prompt leakage for the person using AI. | Building redaction into your own software at scale. |
Bottom line
Private AI and Locke aren't really the same kind of product. Private AI is excellent redaction infrastructure for teams building their own pipelines, with the flexibility of on-premise deployment. Locke is a finished, local-first control for the person actually using an AI tool. If you're building a product that needs redaction, look at Private AI. For the prompts your team types every day, use Locke. Verify current capabilities with each vendor.
