Decision infrastructure for the AI era

The decision authority layer for AI agents

AI agents can analyze and prepare.

Arbitrium — the threshold where automation stops and human authority takes over.

  • They can't sign, attest, or assume legal responsibility.
  • HumanLayer enables an AI agent to obtain a qualified, documented, enforceable human decision — via an API.
Request a demo Read the API docs
1.3B AI agents in use (proj. 2028)
$307B AI software revenue (USD, 2027)
$143B GenAI spending (USD, 2027)
HumanLayer in 60 seconds
1

AI agents prepare decisions

→
2

Authorized humans validate them

→
3

HumanLayer orchestrates and logs and guarantees the deadline

Result: autonomy without loss of accountability.

When AI reaches the regulatory, legal, or ethical boundary, HumanLayer takes over.

Verified Sentinels

Qualifications and authorizations verified

Structured decisions

Standard format, required fields

Full audit trail

AI context, decision, timestamp, identity

Response SLA

Minutes, not days

0B
AI agents in use (proj. 2028)
0B
AI software revenue (proj. 2027)
0B
GenAI spending (proj. 2027)
0B
Total AI spending (proj. 2027)

Projections 2027–2028. Amounts in USD. Sources: IDC (AI software revenue, GenAI/AI spending); Microsoft citing IDC (AI agents).

AI agents fail at the same point: the accountability moment

Today, AI can automate 80–95% of a workflow… But as soon as you need to:

  • validate compliance
  • rule on a gray area
  • approve a spend
  • sign an act
  • issue an enforceable decision

…everything goes back "out of system," manually, without SLA or traceability.

The brake on enterprise adoption isn't execution.
It's the lack of authority.

1
Data collection
2
AI analysis
3
Case preparation
?
Decision & Accountability
The solution

HumanLayer: real-time human decisions, API-native

HumanLayer orchestrates human decisioning exactly where it’s indispensable

1

The AI agent submits

A complete case with context and supporting documents

2

HumanLayer routes

To a qualified Sentinel based on required criteria

3

Arbitrium threshold reached

When the Arbitrium threshold is reached, HumanLayer routes the request to a qualified Sentinel.

4

The Sentinel decides

Structured decision with comments and justification

5

Back to the agent

Decision logged, auditable, ready to execute

Result: AI can complete the workflow end-to-end without breaking compliance.

Designed with Legal, Risk, and Compliance teams.
Who decides?

HumanLayer does not automate the decision

It orchestrates the intervention of an identified, qualified, responsible human decision maker.

Assigned to a real person
Timestamped
Justified
Audited
What HumanLayer prevents

Automating decisions without an identified human accountable for them.

Architecture

The 3 validation levels

Each level maps to a type of responsibility and expertise

1

Compliance validation

Structured checks of rules and policies.

Examples
  • KYC/AML
  • GDPR
  • Spend approval
  • Internal policy compliance
Value: speed + accountability + audit
2

Contextual judgment

When rules aren't enough, an expert decides.

Examples
  • Is a contract clause acceptable?
  • Process exception?
  • An anomaly requiring escalation?
Value: you buy judgment, not time
3

Decision authority

Final act reserved for an authorized person.

Examples
  • Qualified signature
  • Credit decision
  • Medical validation
  • Regulatory approval
Value: assumed accountability + strong barrier to entry
Applications

Use cases

HumanLayer integrates into every sector where accountability matters

Credit & Risk

"AI analyzed 200 credit files. For this customer, the score is in a gray zone (480). Approve the $50,000 loan?"

The AI prepares the complete file with history and recommendation; a certified analyst decides under a 2-hour SLA.

Enforceable decision + audit trail

Legal & M&A

"Out of 500 contracts analyzed, 12 include a change-of-control clause. Is clause §4.2.b acceptable?"

The AI flags critical points; a qualified lawyer validates each risky clause with comments.

Qualified legal opinion

Health & Diagnostics

"The algorithm detects an anomaly on patient #4521's MRI. Tumor probability: 78%. Confirm the diagnosis?"

The AI pre-analyzes imaging; an authorized radiologist validates/adjusts the diagnosis within 30 minutes.

Assumed medical accountability

HR & Hiring

"3 candidates shortlisted by AI. Candidate B has a 2-year gap on their resume. Process exception?"

The AI pre-screens and prepares files; an HR director validates sensitive decisions with justification.

Documented non-discrimination

Industrial Compliance

"Batch #7823 shows a 0.3% deviation from spec. Below the critical threshold but above tolerance. Release the batch?"

The AI detects deviations; an authorized quality manager decides on release.

FDA/ISO traceability

Insurance & Claims

"Auto claim: photos analyzed, statements cross-checked. Timing inconsistency detected. Fraud?"

The AI flags suspicious cases; a mandated expert decides on compensation.

Documented, contestable decision

Spend & Procurement

"SaaS purchase request: $45,000/year. Budget not planned. 3 quotes obtained. Budget exception?"

The AI validates compliance; a CFO approves exceptions according to policy.

Internal control + speed

Cybersecurity & Access

"Admin access request to prod server. Justification: "urgent debug". User: junior dev, 10:30pm. Approve?"

The AI evaluates risk and context; a CISO validates sensitive access in real time.

Zero-trust + audit complet
Differentiation

Why HumanLayer

This is not a task marketplace.
HumanLayer is an accountability infrastructure.

Verified Authority Network

Qualified Sentinels, traceable accreditations

Decision-as-an-API

Endpoints, formats, agent integrations

Audit by Design

Evidence, logs, context, versioning

Guaranteed SLAs

Guaranteed timelines, escalations, fallback

Security

Trust & compliance

Built for Legal, Risk, and Compliance teams

Verified identity

Qualifications and authorizations verified

Audit log

Exportable, full chain of evidence

Advanced policies

Red flags, refus automatique, escalades

Enhanced security

Access control, segmentation

Data minimization

Collection limited to what is strictly necessary

Certifications

SOC2, ISO, RGPD ready

Developers

HumanLayer Systems API

HumanLayer integrates like Stripe or Twilio — but for decisions. Arbitrium is exposed as an escalation state in the system.

POST /decisions Submit a case
GET /decisions/{id} Status & SLA
POST /escalations Escalation / second opinion
POST /proofs Evidence & attestations

Webhooks

decision.completed decision.rejected sla.breached

SDKs & Connectors

Python TypeScript MCP Tools Agent Frameworks
View the API docs →
example.py
# Submit a decision via HumanLayer import humanlayer client = humanlayer.Client(api_key="hl_...") # Create a decision request decision = client.decisions.create( type="compliance_check", level="conformity", context={ "document_id": "doc_abc123", "ai_analysis": analysis_result, "urgency": "standard" }, sla_minutes=30 ) # Wait for the decision result = decision.wait_for_completion() if result.approved: # Continue the workflow process_approval(result.decision)
handle_edge_cases.py
# Handle rejections and escalations result = decision.wait_for_completion() if result.status == "rejected": # Decision rejected by the Sentinel log_rejection( reason=result.rejection_reason, decided_by=result.decider_id, timestamp=result.decided_at ) notify_requester(result) elif result.status == "sla_breached": # SLA breached → automatic escalation escalation = client.escalations.create( decision_id=decision.id, reason="sla_timeout", escalate_to="senior_reviewer" ) result = escalation.wait_for_completion()
Join the network

Are you an expert? Become a HumanLayer Sentinel

Lawyer, physician, analyst, CISO... Monetize your decision authority by validating AI agents' requests. Full flexibility, paid per decision.

Apply as a Sentinel

Application reviewed within 48 hours

Pricing

Decision-based pricing

Not hourly — pay for value, not time

Compliance

Validation of rules and policies

$ 15 / decision
  • Structured verification
  • 30-minute SLA
  • Full audit trail
  • Verified Sentinels
Get started

Expert judgment

When rules are not enough

$ 80 / decision
  • Sector expertise
  • 2-hour SLA
  • Detailed comments
  • Second opinion available
Get started

Authorized authority

Final act by a qualified person

$ 200 / decision
  • Qualified signature
  • Custom SLA
  • Assumed accountability
  • Complete certification
Get started

Enterprise options: premium SLAs, dedicated pools, governance, integrations, reporting.

Get pricing →

Deploy AI agents… without losing control

HumanLayer makes autonomy auditable, accountable, and enforceable.

As AI advances, HumanLayer defines when not to automate.

Request a demo Talk to a compliance expert