50/50
Correct action outcomes in the support benchmark
Let agents do real work.
VerifiedX checks whether a high-impact agent action is justified before it happens.
Protect one high-impact action first. Your runtime, orchestrator, tools, and memory stay the same.
Install in 2 steps
- Get your API key
-
Run the installer
Codex, Cursor, Claude Code.
VERIFIEDX_API_KEY=vxpk_... npx -y @verifiedx-core/sdk@latest install --cwd .$env:VERIFIEDX_API_KEY="vxpk_..."; npx -y @verifiedx-core/sdk@latest install --cwd .
Public proof
Trust should be earned with evals.
People add VerifiedX to systems already in production. So the public proof we lead with is open evals against real workflow classes: support agents, legal agents, and composed systems that can change state.
0/29
Unjustified support actions executed with VerifiedX
21
Workflows completed after VerifiedX replanned the first requested action
LABE
Legal Action Boundary Eval for negotiation, compliance, and review flows
In the support benchmark, baseline completed 22/50 workflows but executed 5/30 unjustified actions. Prompt-only completed 18/50 and executed 1/30. VerifiedX completed 42/50, returned receipts for the rest, and executed 0/29 unjustified actions.
Use VerifiedX when your agents can:
- update records
- write memory
- change workflow state
- send messages
- call webhooks
- take other high-impact actions that affect business
Unjustified actions happen for both adversarial and non-adversarial reasons: content injection (web search results and user-generated inbounds), weak grounding, stale context, drift, overeagerness, and long-term memory poisoning.
What you get
- fewer unjustified actions in production
- agents that get the job done without executing unsupported state changes
- clear decisions your system can act on
What decisions are based on
VerifiedX does not judge an action in isolation.
It looks at:
- what the agent is about to do
- what the agent has already done in this run
- what the agent has ingested into its context from external retrievals like web searches and user-generated inbounds
- the evidence the agent gathered before the action
- upstream context when the run is part of a multi-agent or agent+human system
Works for real agent systems
Built for single-agent, multi-agent, and agent+human systems.
In a single-agent system, that replan loop stays with the same agent.
In a multi-agent or agent+human system, VerifiedX returns the decision in a way that lets the next agent, orchestrator step, or human step take over cleanly.