ServicesPrivate AI applications
Internal AI tools that respect who may see what
A document assistant that answers from your files is only safe if it knows whether the person asking may open those files. We build and deploy internal AI applications on your infrastructure, with permission-aware search, single sign-on and full logging.
- Hosted in your environment
- Answers cite their sources
- Permissions checked on every question
What's our process when a customer asks us to delete their data?
Log the request in the privacy portal the same day, verify the customer's identity, then delete across the CRM, billing and support systems. Legal is copied if there's an open dispute.
Sources: Data subject request procedure v4, Retention scheduleAnswered from 2 documents you can open. 1 more matched but is restricted to Legal, so it wasn't used.
How we build them
Every build includes all of these.
Permission-aware search
Access rights from SharePoint, Google Drive, Confluence or your document system checked at the moment of each question, not copied once at setup.
Single sign-on and roles
Access through your identity provider. Leavers lose access when their account is disabled.
Answers with sources
Every answer cites the documents it came from, so people can check it, and says so when it doesn't know.
Deletion that follows the source
When a document is deleted or its retention period ends, it's removed from the search index too.
Your infrastructure
Application, index and logs in your cloud account, using a private or region-pinned model.
Measured before launch
Tested on a set of real questions your team writes, with accuracy checked by the people who know the answers.
How a question gets answered
question
your documents
inside your environment
GDPR Art. 25
Permissions are checked when the question is asked, against the source system, so a document someone lost access to yesterday is not used today.
What clients ask us for
For when nothing off the shelf fits your data rules.
Policy and knowledge assistants
Staff ask questions in plain words and get answers from the current, approved version of your procedures.
Contract and case review
Clauses, dates and obligations pulled out and compared against your standards, for a lawyer or case handler to confirm.
Client portals
Clients upload, track and get answers about their own matters, and only their own.
Live dashboards
Numbers pulled from your systems into one view, with plain-language summaries for people who don't read dashboards.
Questions
Why not use the assistant built into our office suite?
Sometimes that's the right answer, and we'll say so. A custom build makes sense when the data lives in several systems, when it has to run on a specific model or in a specific place, or when the built-in tool can't enforce your permission model.
Can it make things up?
It can, which is why answers cite their sources, the assistant is told to say when it can't find an answer, and we measure accuracy on your real questions before anyone relies on it.
Who owns the code?
Ownership is set out in the written agreement before the build starts. It runs in your environment either way.
How our live systems handle data
Every case study has a section on how its data is handled: what the AI sees, what it never sees and which rules that maps to.