API-first.
990+ endpoints.
Zero compromises.
Ship AI that actually works. Guardrails built in. 9 providers, one SDK. Evidence generated automatically.
Full API access on every plan · Python & JS SDKs · $10 free API credit
Built for builders
Everything you need. Nothing you don't.
A complete governance and execution layer for AI in production. Ship with confidence on day one.
990+ API endpoints
Full REST API. Everything in the UI is available programmatically. No second-class citizens.
Guardrails as Code
Define governance policies in your codebase. Ori evaluates every AI action. Evidence generated automatically.
9 Providers, One SDK
OpenAI, Anthropic, Groq, Mistral, and more. Unified interface. Intelligent routing. One bill.
Ara Executes for You
AUTO handles the routine. SOFT nudges you. HARD waits for approval. 84 actions, fully programmable.
Webhooks & Streaming
Subscribe to governance events. Trigger workflows from your app. Real-time SSE streaming.
Version Everything
Every workflow versioned. Compare, rollback, branch. CI/CD-friendly from day one.
How it works
One API call. Full governance.
Your app talks to ARQERA. ARQERA handles governance, routing, and evidence. You stay in control.
from arqera import Arqera
client = Arqera(api_key="ak_...")
# Evaluate an AI action with governance
result = client.governance.evaluate(
action="generate_report",
model="gpt-4o",
context={"department": "finance"},
tier="SOFT"
)
# Ara executes with human-in-the-loop
task = client.ara.execute(
command="Summarise Q4 metrics",
tier="AUTO"
)
print(task.status) # "completed"
$ arqera ara execute --command "Deploy staging build"
Evaluating governance policies...
✓ 3/3 policies passed
✓ Tier: HARD — awaiting approval
$ arqera ara approve task_8f3k2
✓ Approved. Deploying to staging...
✓ Build deployed in 42s
✓ Evidence artifact emitted: ev_a91c
SDK
Python. JavaScript. Your language.
Type-safe SDKs that mirror the full API surface. Install, authenticate, ship.
from arqera import Arqera
client = Arqera(api_key="ak_...")
# Governance evaluation
result = client.governance.evaluate(
action="generate"
)import { Arqera } from '@arqera/sdk'
const client = new Arqera({ apiKey: 'ak_...' })
// Governance evaluation
const result = await client.governance.evaluate({
action: 'generate'
});Human-in-the-loop
Three tiers. Total control.
Every AI action is classified, governed, and auditable. You decide what runs free and what needs your sign-off.
Instant. Silent. Done.
Read, search, analyse, monitor. Low-risk actions execute without interruption. Fully auditable after the fact.
{ tier: "AUTO" }
200 completed in 120ms
30-second undo window.
Notifications, suggestions, soft edits. You see what happened and can reverse it instantly via API or UI.
{ tier: "SOFT" }
200 undo window: 30s
Explicit approval. Always.
Deployments, deletions, payments, external comms. Nothing irreversible happens without your explicit API call.
{ reason: "Verified" }
200 approved + executed
Why ARQERA
Built by developers. For developers.
No vendor lock-in
9 AI providers behind one API. Switch models with a config change, not a rewrite. Your governance policies work across all of them.
Compliance from day one
EU AI Act, GDPR, SOC 2 evidence generated automatically. Every AI decision logged, every action auditable. No bolting it on later.
Real-time, not batch
SSE streaming, webhooks, sub-200ms governance evaluation. Your users get instant responses while every action is governed in-line.
Open API, closed loop
Full REST API, Python and JS SDKs, CLI tools. Everything in the UI is available programmatically. CI/CD-native from the start.
Ship in minutes. Scale forever.
Early access available. Cancel anytime. Full API access on every plan.