Skip to content
LargeQuant Runtime Guard

Put a quantitative control plane before the action.

An agent can propose the action. LargeQuant checks value, policy, evidence, uncertainty, liquidity and downside before the decision is executed.

Get API access →Read API documentation
Runtime GuardREQUIRE HUMAN
✓Mandate limitPass
✓Source evidencePass
!Downside liquidityHuman review
✓Calculation evidenceSigned
REST API · MCP · Python · TypeScript

Make quantitative policy executable.

Use assurance primitives from AI agents, enterprise copilots or autonomous systems. Every response binds the input, policy checks, outcome and result into signed evidence.

ALLOWBLOCKCHALLENGEREQUIRE EVIDENCEREQUIRE HUMANRECOMPUTEDEFER
POST /api/v1/guard/check
{
  "action_description": "Transfer operating cash",
  "proposed_value": 4200000,
  "currency": "ZAR",
  "confidence": 0.91,
  "available_liquidity": 18500000,
  "source_evidence": ["LQ-2026-09-001"]
}
01

Policy

Bind action types, value limits, evidence, confidence and review requirements.

02

Preflight

Check the exact proposed action before execution, with idempotency protection.

03

Outcome

Return an enforceable Runtime Guard outcome with explicit reasons.

04

Evidence

Sign the policy snapshot, input, checks and outcome for independent inspection.