Compliance Policy Pack

A verdict your agent can act on — automatically.

Plica doesn't just score a file. It returns a typed, machine-readable enforcement action — approve, route to review, reject, request re-submission — so your workflow or AI agent acts on it without a human re-reading the result.

Try the API — free → See the forensic check Evidence Packet →
Why it matters

A score is not an instruction.

"This file is 55/100" tells an automated workflow nothing. What does it do with 55? The Policy Pack closes that gap: every verdict carries a typed action and the reason codes behind it, so an agent or webhook enforces the right outcome — and suspicious files route to a human instead of clearing silently.

The actions

Four enforceable outcomes.

APPROVE
Proceed
No anomalies across the layers. The file clears and the verdict is recorded with its reason codes.
REVIEW · HITL
Route to human
Ambiguous or conflicting signals. The file goes to the manual compliance queue instead of auto-clearing or auto-rejecting.
REJECT
Deny / escalate
Strong synthetic or manipulation signals. Flagged for rejection in identity, KYC, or claims-evidence workflows.
RESUBMIT
Request original
Sterile or degraded file. Trigger a request for the raw, uncompressed file directly from the source device.
What comes back

Flat, machine-readable, ready to enforce.

// per-file response (illustrative)
{
  "verdict_tier": "suspicious",
  "risk_tags": ["GAN_EMBEDDED", "LAYER_CONFLICT_EMBEDDED"],
  "policy_action": "route_to_hitl",
  "policy_reasons": [
    "authentic scan container conflicts with elevated embedded pixels"
  ],
  "isolate_from_training": true
}
Tags and actions are stable strings — your webhook, n8n flow, or agent maps them to enforcement directly.
How it plugs in

From upload to enforced outcome.

01
Send
File hits the API — image or PDF.
02
Verdict
Forensic layers merge into one tier + reason codes.
03
Policy
Typed action + tags returned in the response.
04
Enforce
Your agent or webhook acts — no human re-read needed.
Honest scope

The action follows the verdict.

The policy action is derived from the verdict tier, not a separate opinion — so the file never carries a "real" engine result and an "approve" action that contradict each other.

A suspicious or conflicted file routes to human review by default. The Policy Pack is built to escalate uncertainty, not to auto-clear it.

Plica is an independent layer. Policy actions feed your existing KYC and case-management stack — they don't replace it.

Wire one verdict into your workflow.

Free trial API · typed policy actions · webhook-ready

Try the API — free →