What happened
On Oct 19, SandboxAQ expanded its Switch AI offering with two new products: Switch Trust (Discover agent/model/tool/MCP inventory; scan code and cloud for vulnerabilities with AI-assisted triage mapped to OWASP agentic categories; Evaluate by attacking agents with prompt injection/jailbreak/data-leakage probes; and Monitor live interactions to detect and block/redact prompt injection, secrets, PII and jailbreaks) and Switch ROI (track per-agent/session spend and outcomes, calculate cost-per-outcome and net ROI, and rank savings recommendations). Both are free for up to five managed agents and work with any vendor's agents.
Why it matters
Switch Trust bundles agent discovery, static scanning, live red-teaming, and runtime enforcement into one vendor-neutral offering, while Switch ROI ties agent security and spend to outcomes — a practical answer to uncontrolled agent sprawl and the failure of many agentic projects to prove value.
Applicability
Enterprises scaling agentic AI across mixed vendors should evaluate Switch Trust for agent/red-team/monitor coverage and Switch ROI to justify agent spend; the free five-agent tier is a low-cost way to pilot on a high-risk workflow.