RakshIQ
Self-hosted PII detection & redaction for AI pipelines
Rakshiq provides a robust, self-hosted system for identifying and redacting Personally Identifiable Information (PII) before it interacts with your internal models.
Key features include:
• 100% offline operation, ensuring no data ever leaves your infrastructure
• Presidio-compatible API for seamless integration with existing setups
• Specialized detection for Indian PII like Aadhaar, PAN, GSTIN, and UPI, including checksum validation
• Multilingual name and place detection supporting both Hindi and English
• Enhanced accuracy with checksums and libphonenumber, significantly reducing false positives
This engine is designed to sit directly on your data's hot path, processing various inputs from chat texts to uploaded documents, call transcripts, and even API keys. It employs a two-stage detection process and offers six anonymization operators (replace, mask, hash, redact, keep, and reversible encrypt), configurable per entity type to match your organization's specific security policies. Rakshiq ensures that sensitive information is meticulously handled without ever being written to disk outside your secure environment. It supports various file types like PDF, DOCX, and TXT, and maintains timing for redacted call transcripts.
Ideal for organizations with stringent data privacy requirements, particularly those handling confidential customer data in diverse formats. It's especially valuable for engineering teams and data privacy officers in financial, healthcare, and e-commerce sectors needing to comply with data protection regulations and secure their internal data pipelines.