Salus

Salus

Guardrails to validate your agent's actions before they execute

Winter 2026ActiveB2BInfrastructureDeveloper ToolsB2BAPIInfrastructureAISan Francisco, CA, USA
Company
https://www.usesalus.ai
Your agent processed a refund without looking up the order ID, costing you thousands. You only found out three hours later from a support ticket. Evals, output scoring, and observability can reduce the likelihood of mistakes like these occurring - but there's no solution that inspects and prevents an action as it’s about to execute. Salus does that. We’ve built an API that wraps around your agent and checks its actions at run time, blocking incorrect ones and providing immediate feedback to guide retries. Kevin and Vedant were roommates at Stanford, where they both studied computer science.

Verdict

High Signal
Market Opportunity
AI agent guardrails and runtime validation is a real and growing need as agentic AI deployments scale across enterprise. The ICP is clear — companies running autonomous agents in production where mistakes have financial consequences (e.g., refund processing, customer ops). B2B infrastructure for AI agents is a large and rapidly expanding TAM driven by the agentic AI wave.
Low Signal
Founder Signal
Both founders are current Stanford undergrads (Kevin Pan: CS, expected graduation unclear but internships suggest ~2026-2027; Vedant Singh: Math & CS, Sep 2023 – Jun 2027, so a sophomore/junior). Kevin's most relevant experience is a 3-month GTM internship at WindBorne Systems and a 6-month analytics internship at TPG — neither is engineering-heavy or AI-relevant. Vedant's LinkedIn shows a 2022 research internship and HackerRank certificates as highlights, with no industry engineering experience. No exits, no shipped products, no relevant industry depth.
Low Signal
Competition
No competitor data was returned, but this space has real incumbents and fast-moving players: Guardrails AI, NeMo Guardrails (NVIDIA), LangSmith (LangChain), Patronus AI, and Aporia all address LLM/agent safety and validation. Large platforms like LangChain and LlamaIndex are adding guardrail features natively. The moat here is unclear and the space is crowded with better-resourced teams.
Low Signal
Product
No named customers, no pricing page, no live demo, no revenue or usage metrics mentioned. The product is described conceptually as an API that wraps agents and validates actions at runtime, but there is zero social proof, no testimonials, and no press coverage. Pure concept stage with a landing page.
OverallC Tier

Salus is attacking a real problem — runtime action validation for AI agents — but everything else is a concern. Both founders are Stanford undergrads with minimal real-world engineering experience; Vedant graduates in 2027 and Kevin's experience is two non-technical internships. There is no visible product traction, no customers, no press, and no evidence of shipped code beyond the concept. The market is legitimate and the insight (pre-execution validation vs. post-hoc observability) is directionally interesting, but the competitive landscape is brutal with NVIDIA, LangChain, and multiple funded startups already in this space. Without a technical moat or early customer evidence, this reads as a smart student project that needs 2-3 more years of founder maturation to be fundable on its own merits.

Active Founders

Kevin Pan
Kevin Pan
Founder

Building to validate your AI agent's actions before they execute.

Vedant Singh
Vedant Singh
Co-founder, CTO

Building to validate your AI agent's actions before they execute AI researcher and formerly @Stanford CS

Salus
Salus
TierC Tier
BatchWinter 2026
Team Size2
StatusActive
LocationSan Francisco, CA, USA