Subjecttreasury@acmeScopetransfer.usdConstraint≤ €10,000 / daySignatures2 of 2Expires2026-06-01T00:00Z
Portable Authorization & Runtime Verification for AI Systems·api.humanos.tech
AI can act.
Make sure it was allowed to.
Allow systems to verify portable authorization before execution and recover missing approval in real time. Ensure AI only acts within approved limits — and generate independently verifiable execution receipts afterwards.
Loved by startups and companies with billions in revenue
Professional servicesPwCBig Four · audit & assurance
Insurance · globalFidelidade7M customers · 14 countriesHealthcareLusíadasNational private hospital networkDiagnosticsJoaquim ChavesMillions of exams / yearBankingnumo.Regulated payment railsHealthtechBloodflowVascular care decisionsClinical softwaremedifyPrescriptions & referralsSafety-criticalCritical SoftwareAerospace · defense · rail
Healthcare techGlinttHospitals & pharmacies, EU-wide
CybersecurityCiberbitPrivileged access · incident response
Industrial opsExtraflowShop-floor sign-offs at scale
OperationsRalioAudit-grade approval workflows
Regulated AI · UKDataWhisperMulti-Agent AI Operating SystemProfessional servicesPwCBig Four · audit & assurance
Insurance · globalFidelidade7M customers · 14 countriesHealthcareLusíadasNational private hospital networkDiagnosticsJoaquim ChavesMillions of exams / yearBankingnumo.Regulated payment railsHealthtechBloodflowVascular care decisionsClinical softwaremedifyPrescriptions & referralsSafety-criticalCritical SoftwareAerospace · defense · rail
Healthcare techGlinttHospitals & pharmacies, EU-wide
CybersecurityCiberbitPrivileged access · incident response
Industrial opsExtraflowShop-floor sign-offs at scale
OperationsRalioAudit-grade approval workflows
Regulated AI · UKDataWhisperMulti-Agent AI Operating SystemThe execution gap
Every AI action must be verifiable before execution and defensible afterwards.
Authorization still lives across emails, chats, PDFs, and disconnected workflows. AI systems execute anyway.
Without verifiable authorization, every automated action becomes a liability.
- 01AI agent executes action.
- 02No verifiable authorization exists.
- 03Compliance blocks rollout.
The authorization layer
Human intent becomes runtime authorization.
Humanos turns approvals, contracts, policies, signatures, and delegated authority into runtime authorization systems verify before execution.
Let systems autonomously verify whether actions are allowed before execution.
Existing approval
Authorize treasury transfers up to €10,000 / day to AWS until June 1.
✔ Signed
Live mandateMandate0xA13F…E2C9
Verified before execution
- AI Agent
- Bank
- ERP
- Auditor
- Partner API
Built to operate within existing regulatory frameworks, enabling defensible and compliant autonomous execution.
- ISO 27001
- DORA
- NIS2
- EU AI Act
- GDPR
- HIPAA
- eIDAS 2.0
01 / Portable authorization
Portable runtime authorization.
Turn existing approvals and policies — collected through Humanos or external systems — into runtime authorization AI systems verify before acting.
SignatureConsentKYCContractSMSWhatsAppFormPolicyPaymentVoiceEmailSlackPDFWebhookWallet PassOAuthIdentity CheckApproval FlowDocumentInvoice
Mandatesigned
scopetreasury.payments
limit€10k / day
vendorAWS
subjectacme-treasury
expires2026-12-31
Verified before execution
02 / Cross-system propagation
Reusable across systems.
Runtime authorization propagates across APIs, agents, workflows, and external systems — allowing every participant to verify what actions are actually allowed before execution.
Mandateauth_pK7g3xN9
scopetreasury.payments
limit€10k / day
- Bank API
- ERP
- Treasury agent
03 / Verify before execution
Verify before execution.
Every action passes through deterministic checks — scope, freshness, identity, revocation — before it touches a system of record. No verification, no execution.
01
Action
02
humanos.verify(action)
03
Authorized
04
Execute
humanos.verify( action: "wire.transfer", amount: 4200, vendor: "AWS" })
AUTHORIZEDWithin approved runtime scope. Execution proceeds.
COLLECT_MISSINGOutside approved scope. Authorization recovered live.
REJECTEDConstraints violated. Execution blocked before runtime.
04 / Dynamic recovery
Recover authorization dynamically.
When execution falls outside approved limits, Humanos pauses the action, collects what's missing in real time, and resumes automatically once authorization is verified.
01AI agent attempts paymentaction
02Amount exceeds approved limitpaused
03CFO approval requested liverecovering
04Approval verifiedauthorized
05Payment execution resumesresumed
05 / Execution receipts
Portable execution receipts.
Every action emits independently verifiable proof — anchored, signed, and re-verifiable by any party that needs to attest to what happened.
Execution receiptEd25519 · verified
actionwire.transfer
amount€4,200
auth_idauth_pK7g3xN9
timestamp2026-04-12T14:23:08Z
signature0x4f8a…c2e1
hashsha256:9f3c…8b41
Independently verifiable by
AuditorRegulatorCounterpartyBankInsuranceCourtPartner systemBlockchain exploreranyone with the public key
No callback to Humanos required.
Network & trust
Every authorized action becomes independently verifiable.
Existing permissions, approvals and policies become reusable runtime authorization across systems, APIs and autonomous agents.
Reuse · 01
Reuse portable authorization across systems
one mandate, accepted everywhere
Identity · 02
Verify identity and signer authenticity
signers verified cryptographically
Policy · 03
Policy changes without redeploying
rules propagate live through mandates
Audit · 04
Independently verifiable audit trails
every action anchored, re-verifiable forever
Portable · 05
Portable authorization across organizations
authority that travels with the actor
Runtime · 06
Runtime enforcement for autonomous systems
agents that cannot act outside their mandate
Humanos separates authorization from execution.
Real customers already verifying before execution
Numo agents settled treasury payments in real time, nothing executed outside approved limits.
How it works
- 01Every transaction calls
verify()before execution. - 02Within limits executes instantly.
- 03Out of limits approval is requested.
- 04Every decision produces a signed execution receipt.
await humanos.verify({ amount, counterparty, mandate })
Integration
Integrates into any execution path.
Your systems already decide who can act. Humanos allows external systems and autonomous workflows to verify what they were actually allowed to do before execution.
Drop verify() into agents, workflows, APIs, ERPs, CRMs, or internal tools.
verify()blockedapprovalauthorized
SDK · MCP · API · Agents
Add verify() to any system.
Verify actions at runtime. Block, request approval, or resume execution — all in one call.
Run verify() →Triggerverify()Action
Workflows
Control actions in automation flows.
Drop a verify step into Zapier, n8n, or skills before actions execute.
Open template →Existing stack
Works with your current systems.
Verify actions in front of ERP, CRM, payments, or internal tools. No rewrites required.
Test with your stack →FAQ
Authorization, as infrastructure you call — not a feature you build.
Does Humanos replace our IAM or policy engine?
No. Existing systems continue deciding who can act internally. Humanos standardizes authorization into portable runtime infrastructure external systems and autonomous workflows can verify before execution.
Do we need to replace our existing authorization system?
No. Humanos works alongside existing IAM, RBAC, OAuth, approval workflows, ERP permissions and policy systems. Existing authorization becomes portable and externally verifiable across systems before execution.
Why not keep authorization entirely inside our application?
Internal authorization works inside the application that created it. Autonomous systems increasingly operate across APIs, workflows, organizations and counterparties. Humanos makes authorization portable and independently verifiable outside the originating system.
Can external systems verify authorization without trusting our application?
Yes. Humanos turns approvals and policies into machine-verifiable mandates external systems can verify independently before execution.
Why not build authorization infrastructure in-house?
Humanos replaces fragmented approval collection, runtime verification, escalation flows, audit systems, and compliance logic with one runtime API. Instead of building separate approval workflows, policy engines, audit logs, and verification layers internally, systems call Humanos before execution.
How is this different from human-in-the-loop?
Humanos verifies whether the exact action being executed falls within the scope, limits, counterparties, expiration rules, and conditions previously authorized. This transforms approvals and policies from static documentation into machine-verifiable runtime authorization.
How does this reduce hallucinations and unauthorized actions?
AI systems no longer rely solely on prompts, assumptions, or loosely enforced business logic. Before execution, Humanos checks whether the requested action is actually authorized, preventing systems from operating outside approved boundaries.
What happens when authorization rules change?
Authorization becomes externalized runtime infrastructure instead of hardcoded application logic. Limits, permissions, counterparties, expiration dates, delegation rights, and policies can be updated, revoked, or replaced without redesigning workflows or redeploying systems.
Can the same authorization work across multiple systems?
Approvals collected once can be verified across ERPs, payment systems, AI agents, CRMs, treasury systems, partner organizations, and workflows. Authorization becomes portable infrastructure instead of being trapped inside individual applications.
What happens when an action falls outside approved scope?
Humanos pauses execution, requests approval dynamically through any channel, updates authorization in real time, and resumes execution automatically once approval is granted.
How can every execution be verified afterwards?
Every authorized action produces a signed, portable execution receipt containing who authorized the action, under which constraints, the exact payload executed, timestamps, and verification data. These receipts can be independently verified outside Humanos.
Can auditors verify without our internal logs?
Auditors, regulators, banks, insurers, counterparties, courts, and external systems can verify what was approved and executed without relying on internal logs or trusting a single organization's infrastructure.
How does this affect regulated deployments?
Organizations can deploy autonomous systems into finance, healthcare, insurance, ERP operations, and regulated workflows without rebuilding governance, approval, and audit infrastructure internally from scratch.
Can multiple organizations share authorization?
Different organizations and systems can verify the same authorization object independently, enabling interoperable trust between enterprises, AI agents, partners, platforms, and external systems.
Why not embed this directly in our application code?
Approval enforcement moves out of scattered workflow code and into a shared runtime verification layer. Systems no longer need to embed custom authorization logic into every workflow, service, or AI agent individually.
Verify before execution
Before systems act, make sure they're allowed to.
Authorization collected once. Verified before execution. Proven on demand.
Built for agents, payments, ERP, workflows, and any system executing real actions.