Privacy

PII candidate review and redaction

Review deterministic PII candidates individually or in a batch, apply per-kind replacement rules, and undo the latest batch through a new revision.

Implemented task

Public task

Open source actions and choose Add source.

Expected outcome: A validated source enters the durable revision and processing pipeline.

ApplicationRequires Sign-in, Existing project

Declared product contract

Inputs and task boundary

Inputs

  • A readable text transcript
  • Deterministic email, phone, URL, or IP-address candidates

Input constraints

  • Semantic names, addresses, health data, and organization-specific classes require another detector or manual review.

User actions

  1. Open PII review for a source.
  2. Inspect candidates and choose per-kind replacement rules.
  3. Preview and commit one versioned batch or compensate the latest batch.

Outputs

  • A new transcript revision containing the explicitly reviewed replacement batch.

Decision boundary

Where it fits

  • Reviewing the currently declared deterministic identifier patterns before analysis or sharing.

Outside the boundary

Where it does not fit

  • A guarantee that all sensitive or identifying content is detected.

Current scope

Limitations

  • The deterministic scanner currently covers email, phone, URL, and IP address patterns; semantic names, addresses, health data, and organization-specific classes require another detector or manual review.

Verification

Product proof

journey

proof.m1-material-import

Open artifact
Verified
Review due
Expires
  • Eight public fixture formats enter the real source and revision pipeline.
  • Exact locators and manual coding persist across refresh.
  • Withdrawal previews impact and erases live database and stored objects.

Failure boundary: The proof does not claim M3 typed cases and variables or frame-level video analysis.

OpenVerbatim is an open-source (Apache-2.0) qualitative data analysis platform for coding and analyzing interview transcripts. AI-suggested codes stay marked as suggestions until a human reviewer confirms or rejects them, and every decision is kept in an audit trail. The full feature set is available when self-hosted; there is no paid feature wall.