Execution authorization — deployment & integration
Subtitle: Runtime verification, fail-closed enforcement, and signed runtime proof before operational state changes run.
Executive summary
AI systems are moving from generating text to triggering operational actions. BiDigest inserts a deterministic authorization checkpoint between AI workflows and systems of record. Before payments, CRM updates, claims actions, or outbound communications execute, the orchestration layer calls the authorization verify API and receives a hard runtime verdict plus a verification receipt.
Product category: execution authorization. Stable v1 path: POST /api/v1/admissibility/verify — a versioned alias (e.g. /api/v1/authorize/verify) may ship later without breaking this contract.
The authorization gateway is a strict, fail-closed checkpoint for agentic workflows at the orchestration layer.
Phase 1 — Provision approved authority
Before production traffic, register the domain and regulatory identifiers your agents must respect.
- Map verified corporate domain + regulatory IDs (e.g. SEC, FINRA, NPI, LEI) to your tenant
- Produces a machine-readable approved authority source (provisioned allow-list) agents verify against
Phase 2 — Route workflows through verify
The authorization gateway is a strict, fail-closed checkpoint for agentic workflows.
- Integrate LangChain, Copilot Studio, AWS Bedrock, n8n, Make, or custom orchestration
- Call POST /api/v1/admissibility/verify before CRM writes, payments, refunds, or outbound sends
- Structured payload only — not a full chat transcript
Phase 3 — Branch on verdict and record proof
Resolution is deterministic at this layer — no generative step in the gateway.
- APPROVED — proceed to the downstream system with receipt_id attached
- REVIEW_REQUIRED — pause automation; route to human approval (Slack, email, ticketing)
- REJECTED / fail-closed — stop the binding action; signed execution record documents the denial
APPROVED — proceed with receipt_id attached. REVIEW_REQUIRED — pause for human approval. REJECTED or fail-closed — stop the binding action; signed execution record documents the denial (e.g. execution drift — unverified sources vs provisioned authority; API reason codes are stable machine identifiers).
Payload architecture
Endpoint: POST https://bidigest.com/api/v1/admissibility/verify
Authentication: API key via existing v1 middleware (Authorization: Bearer …).
Request body (example)
{
"entity_anchor": {
"domain": "client-domain.com",
"regulatory_id": "SEC-801-12345"
},
"agent_context": {
"orchestration_id": "prod-langchain-01",
"proposed_action": "transmit_client_advisory",
"prompt_hash": "a1b2c3d4e5f6g7h8i9j0"
},
"payload_to_verify": {
"synthesized_claims": [
"Our firm manages $500M in assets."
],
"shadow_citations": [
"https://client-domain.com/q3-report.pdf"
]
},
"jurisdiction_context": {
"region_code": "eu-central",
"skb_profile_ids": ["eu-ai-act"]
}
}JSON uses stable v1 wire names. Operational mapping: shadow_citations → proposed sources pending execution; decision_receipt → signed verification record (compact JWS); admissibility_status → runtime authorization verdict. Reason codes (e.g. GROUND_TRUTH_VERIFIED) are machine identifiers — explain them as authority / execution drift in runbooks.
Optional — multi-region routing: jurisdiction_context.region_code binds the request to your regional partition (example eu-central); skb_profile_ids stacks regulatory profiles (e.g. EU AI Act slice).
Response — fail-closed (403)
{
"admissibility_status": "DENIED",
"reason_code": "NARRATIVE_DRIFT_DETECTED",
"failed_citations": ["internal_doc_id_9921"],
"decision_receipt": {
"timestamp": "2026-03-23T10:40:27Z",
"receipt_id": "rec_fail_8847xyz",
"signature": "<compact JWS>"
}
}Response — approved (200)
{
"admissibility_status": "APPROVED",
"reason_code": "GROUND_TRUTH_VERIFIED",
"decision_receipt": {
"timestamp": "2026-03-23T10:40:27Z",
"receipt_id": "rec_pass_1122abc",
"evidence_hash": "<sha256>",
"signature": "<compact JWS>"
}
}Integration FAQ & security posture
1. Latency overhead
The gateway does not generate tokens or ingest the full context window; it performs deterministic authorization checks against your provisioned authority source. Overhead is typically single-digit milliseconds plus network — similar to a JWT validation round-trip.
2. Data privacy / PII
The contract is structural: entity anchor, cryptographic hash of the prompt (receipt binding), and the action context pending execution — not full conversational history. Final handling is covered under your enterprise DPA.
3. Availability & fail-closed behavior
If no signed verification record returns within your orchestration timeout, route to a deterministic fallback (static approved response, degraded mode, or human-in-the-loop). You control uptime without inheriting unverified AI liability.
Next step
To provision staging and receive API credentials, request a scoped authorization assessment at https://bidigest.com/governance/admissibility-vault/intake
© 2026 BiDigest · Confidential · bidigest.com/governance/architecture