Data Privacy Filter
A data privacy filter sits between your people and any AI model, stripping personal information out of prompts and documents before anything leaves your building. The cleaned text then goes to a local model or, where the work genuinely needs one, to a commercial model — with the identifiers still on your side of the wall. It is the component that makes AI usable on records you are legally responsible for.
Where this one earns its keep.
Every sector we work in has a version of this problem. These four feel it hardest, and they are where we have built it most often.
What one pasted document can cost.
No invented example here. The method is below; the numbers are yours to put into it.
What to count. documents a month that currently cannot go near a model × the minutes each one costs to handle by hand. That is the arithmetic, and it is yours to run — we would rather hand you the method than a number invented for a page.
The ROI calculator models the same shape across a whole team, and its assumptions are published in full underneath it. The number that actually matters comes out of an AI Readiness Audit, which measures your workflows instead of averaging them.
Discover, build, evolve.
The same three phases as every other engagement. Nothing about this build gets its own process.
Discover. We walk the workflow as it runs today — who touches it, where it stalls, what it costs when it slips. That is the AI Readiness Audit, and it ends in a written document you own.
Build. A fixed quote in writing before work begins, then the system goes into your real operation rather than a demo environment. A person stays in the loop on anything a customer sees.
Evolve. Once it is live it needs watching: what it handles, what it escalates, what changed in your business since. That is Managed Services, and it is optional.
Delivered through AI Implementation Sprint.
This build has no price of its own. It is scoped and delivered inside a published service, at the published price.
An AI Readiness Audit runs $3,500–$8,500 and tells you whether this build is the right first move. The build itself is delivered in AI Implementation Sprint. $18,000–$60,000, quoted fixed in writing before work begins.
Data Privacy Filter, asked plainly.
01What counts as personal information here?
Names, contact details, account and file numbers, and whatever else your sector treats as identifying — which is a scoping conversation, because a lender, a clinic, and a contractor do not have the same list.
02Does the filter slow things down?
Not in a way anyone notices. It runs on the same machine as the model, so the work stays inside your building rather than making a round trip.
03Can we prove it worked?
That is what the audit trail is for. Every request through the filter is logged, so the question 'did that document leave the building' has a recorded answer rather than an assurance.
04What does it cost?
It is delivered through an AI Implementation Sprint, quoted fixed in writing before work begins. It is normally scoped alongside the private server rather than on its own.
30 minutes. A straight answer.
A senior advisor walks your workflows and tells you whether this build is worth doing — including when the answer is not yet.