Fast, dependency-free Go library that replaces sensitive data with realistic fakes while preserving structure and semantic meaning. Fully deterministic (no LLM calls) — built for safe testing, demos, and AI pipelines.
Comments
addieg•1h ago
We’re building a zero-trust, on-demand data fabric that runs in your VPC and spins up AI-ready views across silos, every query or LLM sees only the slice it should.
One key challenge is handling PII: don’t expose it to the LLM, but don’t lose the meaning either. We couldn’t find a good solution in Go, so we built one: fast, deterministic, and no external dependencies.
The lib preserves semantic structure — think emails that look like emails, ZIP codes/phone numbers that match state, and names that remain gendered. Curious what other use cases folks might see for this!
addieg•1h ago
One key challenge is handling PII: don’t expose it to the LLM, but don’t lose the meaning either. We couldn’t find a good solution in Go, so we built one: fast, deterministic, and no external dependencies.
The lib preserves semantic structure — think emails that look like emails, ZIP codes/phone numbers that match state, and names that remain gendered. Curious what other use cases folks might see for this!