Two kinds of product take personal details out of documents. One is a component a developer builds into a pipeline. The other is a service that hands back redacted files and the evidence for whoever signs off. The right one depends on who does the work and who has to answer for the result.
What Limina does
Limina is the new name of Private AI. Its own announcement says the API, the features and the deployment did not change. It describes its product as software that removes personal data from text, images, audio and documents, delivered as a container that runs in your own cloud or on your own servers.
It is broad. It reads PDFs, Office files, images and audio, returns a redacted audio file with its transcript, and covers dozens of languages. The container contacts the vendor for authentication and usage reporting. Its pricing page lists a free starter tier by daily API calls, and batch and enterprise plans by quote.
A component and a service
Limina is built for engineers. You call it from your code, and what comes back is processed text and a list of what it found. Its documentation describes no human review step and no report written for a privacy office or counsel, so the checking and the sign-off are yours to design.
Side by side
| Question | Limina | PHI Mask |
|---|---|---|
| Who it is for | Engineers building de-identification into a product or pipeline | The people who hold the documents, and whoever signs off on releasing them |
| What comes back | Processed text and a list of the entities found | Masked copies of the documents; for a batch, a release report and a reviewed sample |
| How it finds values | Models that read context, with allow and block lists | Pattern matching and OCR, with no LLM in a standard batch |
| Where it runs | A container in your environment that reports usage to the vendor | The browser tab you opened; a batch runs offline in your environment |
| Published price | A free starter tier; batch and enterprise by quote | A price by the page, printed on the page |
| Before the first file | An account and a container to deploy | Open a link |
Where Limina is the better choice
If you need audio, many languages, or an API that screens text in real time inside your own software, Limina covers ground PHI Mask does not. It also offers synthetic replacement values, where PHI Mask writes placeholders. A team with engineers and a product to build is its natural buyer.
Where PHI Mask fits
PHI Mask is for the people who hold the files. One document is masked in the browser, with each value marked for you to approve. A whole document set goes through PHI Mask Batch, which runs offline in your environment or with our team, and returns redacted copies, a release report of counts and hashes, and a sample that people have read. Its detection is pattern matching and OCR, so the same input gives the same output on every run.
Review before sharing.
Sources and review notes
Every statement about Limina above comes from its own current pages, reviewed on the date shown at the top of this page: its home page, the rename announcement, supported files, audio processing, languages, the documentation FAQ and pricing. Where its pages are silent on a capability, this page says so and does not guess. Products in this category change quickly, which is why this comparison carries a recheck date.
Common questions
- Is Limina the same product as Private AI?
- Yes. Private AI renamed itself Limina, and its announcement says the API, the features and the deployment did not change.
- Which one handles audio and other languages?
- Limina. It documents audio files and dozens of languages. PHI Mask works on text, PDFs, scans, images and spreadsheets.
- Which one gives a privacy office something to sign?
- PHI Mask Batch returns a release report of counts and hashes and a sample that people have read. Limina's documentation describes an API that returns processed text and the entities found.
- Does either one use an LLM?
- A standard PHI Mask batch uses pattern matching and OCR, with no LLM or generative AI. Limina describes models that read context.