Blackline is a privacy-first document redaction and substitution platform. It combines AI-powered entity recognition, human-in-the-loop verification, and cryptographic proofs of integrity to enable safe data sharing and AI ingestion.
Built for legal teams, compliance officers, and privacy engineers who need precise, auditable control over what gets disclosed and what stays protected.
- Redact: Completely obscure sensitive information (e.g., "The budget is
[REDACTED]"). Best for public release or legal discovery.
- Substitute: Replace sensitive entities with context-preserving placeholders (e.g., "John Doe" →
[PERSON_1]). Best for Enterprise AI reuse, allowing LLMs to understand relationships and context without accessing actual PII.
Blackline uses a Dual-Path AI Engine (powered by Google Gemini) where user instructions take priority over generic classification. However, because disclosure is high-stakes, the platform enforces a Human-in-the-Loop workflow where every AI suggestion must be approved before the final document is generated.
Safely feed internal documents, transcripts, and reports into LLMs. By using Substitution, you preserve the structural utility of the data for the AI while ensuring no sensitive customer or corporate data leaves your controlled environment.
Streamline the process of preparing documents for FOIA requests, court filings, or general public release. The Manifest provides an auditable trail of what was changed, by whom, and when.
- Instruction-First AI — Tell the AI exactly what to look for (e.g., "Hide all project names and financial figures").
- Manual Overrides — Select any text span in the original document to create custom redactions instantly.
- Bulk Workflow — Efficiently review AI suggestions with bulk accept/reject and quick keyboard shortcuts.
- Native PDF & DOCX Support — Direct ingestion of formatted documents into the AI engine.
- Cryptographic Manifest — Generates a SHA-256 hash-chain manifest linking the original document, the redacted version, and the reviewer's signature.
- Midnight Wallet Integration — Connect a Midnight Lace or compatible wallet to securely sign disclosure attestations, paving the way for Zero-Knowledge proofs of redaction.
- Frontend: React 19, Tailwind CSS v4, Motion (Framer Motion), Lucide Icons
- Backend: Express, Google Gemini AI (
@google/genai) - Blockchain: Midnight Network SDK (Preview)
- Build: Vite 6, TypeScript 5.8
- Node.js v18+
- A Gemini API key
-
Clone the repository
git clone https://github.com/eliasasbrede/Blackline.git cd Blackline -
Install dependencies
npm install
-
Configure Environment
cp .env.example .env # Edit .env and add your GEMINI_API_KEY -
Launch Local Servers
npm run dev:all
Access the app at http://localhost:3000.
- Native PDF/DOCX Support: Direct ingestion of formatted documents.
- On-Chain Persistence: Publishing redaction manifests to the Midnight network for permanent, trustless verification.
- Template Library: Pre-set redaction rules for GDPR, HIPAA, and CCPA.
- ZKP Integration: Verified "Proof of Redaction" where a third party can verify a document was redacted according to specific rules without seeing the original data.
- On-Chain Deployment: The Midnight wallet integration supports local cryptographic signatures. Full on-chain contract persistence is deferred to a future phase.
- File Size: Optimised for documents up to 50MB. Large format PDFs may take slightly longer to parse.
MIT — © 2026 Blackline Protocol.