Human🍁Exe API

The Human-Link.

Turbocharge your AI workflow. Already have an AI API key? You're in the right place.

The Governance Layer for AI That Thinks — structural, provider-agnostic, session-aware.

Human.Exe provides a structural AI governance layer you deploy between your application and any AI provider. Your model. Our governance. Every decision traceable.

Most AI tools give you a response. Human.Exe governs a decision. The difference is architectural, not cosmetic.

Quickstart · PREVIEW

When the API launches, get your first governed inference in under five minutes.

npm install @human-exe/governance
# or
pnpm add @human-exe/governance

import { GovernanceLayer } from '@human-exe/governance';

const gov = new GovernanceLayer({
  apiKey: process.env.HUMAN_EXE_API_KEY,
  provider: 'anthropic',           // your provider
  providerKey: process.env.ANTHROPIC_API_KEY,
  tier: 'citizen',
  audit: true,                      // structured audit trail
  coherency: true,                  // session coherency tracking
});

const result = await gov.infer({
  sessionId: 'gs-demo-001',
  prompt: 'Analyse this quarterly report for risk indicators.',
  context: { domain: 'financial-analysis' },
});

console.log(result.response);          // governed model response
console.log(result.governance);         // audit trail object
console.log(result.governance.session_coherency);  // 0.0 – 1.0

Authentication · PREVIEW

Every request to the Human.Exe API will require two credentials:

POST https://api.human-exe.ca/v1/govern
Authorization: Bearer hxe_your_api_key
Content-Type: application/json
X-Provider-Key: sk-your_provider_key

The Governance Layer

Existing governance tools audit what AI does after deployment. Human.Exe embeds governance into the reasoning process before the response is generated.

The governance layer intercepts every inference request, applies structural governance — scope validation, drift detection, complexity routing — and returns the model response alongside a complete audit trail object. The developer sees inputs and governed outputs. The internals are opaque by design.

Governed Sessions

A single AI response can look great. A session — 20 turns in, with context accumulated, contradictions built up, scope drifting — is where ungoverned AI falls apart. Coherence isn't a feature. It's an architectural property.

A governed session is a sequence of AI interactions managed by the governance layer. Coherency is tracked, drift is flagged, and the audit trail is maintained across every turn — not reconstructed after the fact.

// Turn 1
const t1 = await gov.infer({
  sessionId: 'gs-risk-analysis-001',
  prompt: 'Identify the top 3 risk factors in this dataset.',
  context: { domain: 'risk-assessment' },
});

// Turn 2 — governance tracks coherency across turns
const t2 = await gov.infer({
  sessionId: 'gs-risk-analysis-001',
  prompt: 'Now cross-reference those risks against the Q3 report.',
});

// Coherency monitored across the session
console.log(t2.governance.session_coherency);  // 0.94
console.log(t2.governance.coherency_delta);    // -0.02 (slight drift)
console.log(t2.governance.drift_flags);        // []

Audit Trail

Every critical system produces an audit trail. Medical devices. Financial transactions. Legal records. AI systems making decisions that affect people should produce one too. Human.Exe makes the audit trail a structural output of the governance process — not a logging add-on.

{
  "response": "Based on the dataset analysis, the three primary risk...",
  "governance": {
    "session_id": "gs-risk-analysis-001",
    "request_id": "req-a7f3c901",
    "model_used": "claude-3-5-sonnet",
    "tier_routed": "citizen",
    "coherency_delta": +0.03,
    "session_coherency": 0.91,
    "audit_hash": "sha256:e3b0c44298fc1c149afbf4c8996fb924...",
    "governance_applied": [
      "scope_validation",
      "drift_detection",
      "complexity_routing",
      "audit_logging"
    ],
    "confidence": 0.87,
    "timestamp_utc": "2026-03-27T14:32:01.441Z"
  }
}

Coherency Tracking

A session-level metric that measures how structurally consistent responses are across turns. Flags drift, contradiction, scope violation, and sycophantic pattern collapse.

Coherency tracking is structural — it operates on the governance architecture itself, not as a heuristic bolted onto outputs. The difference between measuring coherency after inference and embedding it in the governance process is the difference between a dashboard and an architecture.

Governed Sparsity Routing

Most AI applications route every request to the most expensive model available. That's architecturally naive. Governed sparsity routing sends 95%+ of requests to appropriately-scoped models based on complexity — producing 10–20× cost advantage at scale.

Don't route simple tasks to Sovereign-tier models. The governance layer auto-routes by default, but you can override with explicit tier preferences per request.

const result = await gov.infer({
  sessionId: 'gs-demo-002',
  prompt: 'Summarise this paragraph.',
  routing: { preferTier: 'observer' },  // force lightweight model
});

Response Format

Every governed API call returns the standard model response plus a governance object. This is not optional — governance is structural, not an add-on.

Approved Models

The following models have been evaluated against Human.Exe structural governance criteria. They are recommended for use through the governance layer. Performance under governance differs from raw API performance — governed sessions produce more coherent, auditable, and cost-efficient results.

Efficiency Best Practices

The governance layer is designed to reduce cost and increase reliability. These practices maximise both.

Governance Layer, Not Intelligence

Let the industry name their capability threshold whatever they want — AGI, Advanced Intelligence, frontier models. That's not our fight. Human.Exe is not a model. We are not an intelligence system. We are the governance layer that sits between users and whatever intelligence system you choose to deploy.

The problem we're solving isn't that models aren't capable enough. The problem is that end-user behavior is unpredictable — and no model, regardless of capability, is structurally safe without a governance layer enforcing bounds. Users drift. Users probe limits. Users behave in ways no training run anticipated. The governance layer is what closes that gap.

Cognitive Benchmarks

Human.Exe maintains an ongoing agent cognitive benchmark programme — structured tests across 5+ major model families measuring reasoning coherency, session drift resistance, adversarial stability, and structural consistency across multi-turn interactions.

Pricing

Phase 2 — Observer, Citizen ($8/mo), and Scholar ($15/mo) are available in early access. Higher tiers open by phase.

Observer gives you the real governance layer, real audit trail, and the full content catalogue. That’s the baseline.

Annual and multi-year plans will be available for Citizen and above when tiers open — save up to 30% with a 3-year commitment.

Engagement Briefs

For parties evaluating Human.Exe at the investment, government, or strategic level.

Observer is in early access — no brief required to access the governance layer. Paid tiers open by phase. These briefs exist for a different conversation: due diligence, institutional evaluation, partnership exploration, and financial strategy. If you're here to build, start on Observer and you'll see the upgrade path clearly. If you're here to understand what you're looking at — request a brief.

AGI R&D

From the AGI R&D Division @Human.EXE

Deep-form content on structural AI governance. No fluff. Respects your intelligence.