Skip to content
Just Healthcare

Clinical privacy intelligence

Dottie™

Big AI. Smaller data footprint.

Dottie™ is a proprietary privacy layer designed to detect, redact, and obfuscate patient identifiers before clinical information reaches external AI models.

justprivacy.health/redactLive
Clinical excerpt
Redaction modePHIPHI + PII
Patient
Maya Bennett[REDACTED]
PHI
DOB
04/12/1973[REDACTED]
PHI
MRN
0049182[REDACTED]
PHI
Phone
202-555-0147[REDACTED]
PHI
Physician
Dr. Elena Park[REDACTED]
KEPT
Facility
Riverside Cancer Center[REDACTED]
KEPT
City
Arlington, VA[REDACTED]
KEPT
Diagnosis
Early-stage breast cancer
Receptors
ER+ / HER2−
Plan
Partial breast irradiation
Scanning for identifiers. Nothing leaves until they are out.
Round tripdottie-mini · WebGPU
  1. 1Detect4 patient identifiers found
  2. 2RedactTokens keep their shape
  3. 3Your AI worksClinical context only
  4. 4Verify0 identifiers echoed back
  5. 5Re-injectReal values restored exactly
Outbound to AIHeld
Retention 0 · In memoryFail closed

What it solves

The models move fast. Privacy should keep up.

Healthcare AI builds on enterprise foundation models from companies such as OpenAI and Anthropic. BAAs, HIPAA/HITECH policies, and enterprise model terms establish important protections; Dottie™ adds a practical control inside the workflow. Identifiers come out before the text is sent and go back in when the work comes home, so the model sees the clinical picture without the patient's identity.

  • Detect, redact, obfuscate

    Names, dates, record numbers, and contact details are found and replaced before transmission.

  • Keep the clinical picture

    Diagnosis, receptors, plan, and timeline stay intact, so the AI's work stays useful.

  • Nothing kept

    Text is processed in memory and released. There is no storage setting to switch on.

A little genius. A lot of discretion.

Less identity. More clinical context.

Protect · Dottie™

How it works

A smarter point of protection. Before transmission.

Identifiers come out before your AI sees the text and go back in, exactly, when the work comes home.

  1. Step 01

    Detect

    Rules and the model work together to find identifiers across the documents.

    Names · dates · MRNs
  2. Step 02

    Redact

    Identifiers become tokens or stand-ins that keep their shape, so dates stay on the timeline.

    Tokens · stand-ins
  3. Step 03

    Your AI works

    The model drafts, summarizes, or reasons over the clinical context, without the identity.

    Clinical context only
  4. Step 04

    Verify and re-inject

    The output is checked for any repeated identifier, then the real values go back exactly.

    Leak check · re-identification

Fail closed

If Dottie™ can't safely remove an identifier, nothing is sent.

Every request is checked before it leaves and again when the answer returns. Zero is the only retention setting. Clinical text is handled in memory for the moment it takes to protect it, then released, and it is never used to train Dottie™ or any downstream model. The key that restores identities is encrypted and held by you, or kept only in your browser's memory.

  • Processed in memory
  • Never used for training
  • Fail closed

What your team gets

Serious about privacy. Delightfully Dottie.

A privacy layer that sits in front of every external model, keeps the clinical story, and keeps nothing else.

  • 01

    PHI and PHI + PII modes

    Remove patient identifiers only, or every person, organization, and location too.

  • 02

    Run it where you choose

    dottie-mini in the browser through WebGPU, or dottie-max on a dedicated BAA-covered server.

  • 03

    Zero retention by design

    No retention tiers, no archive toggle, no PHI in logs, no copy of your data.

  • 04

    Re-identification stays with you

    The vault that restores identities is encrypted and bound to your organization, or held only in browser memory.

Start with the problem that hurts most

Show us where your team is losing time.

We will show you which product fits the work, what your team would see, and how the solution can support the way your organization operates today.