Company Page

Salus

Founders

Salus builds runtime guardrails that validate AI agent actions before they execute. The platform wraps around AI agents to check actions at runtime, blocking incorrect ones and providing immediate feedback to guide retries. Salus maintains an evidence cache for each run and validates proposed actions against policy constraints written in YAML, markdown, or plain English. The platform includes features for PII detection, budget and loop protection, idempotency, human-in-the-loop escalation, and content moderation. Salus has demonstrated strong results on industry benchmarks, reducing AI agent misalignment by 52% on average across 12 frontier models.

Company Founded in 2026