Developers · Quickstart

Up and running in 5 minutes.

From zero to your first governed AI spend intent — with a live policy decision, an immutable audit entry, and full compliance coverage.

🕐 ~5 min setup time
📄 6 steps to first decision
🌟 Free starter plan
1
Create your account

Sign up at the AegisOS Comply dashboard. During onboarding you will name your organisation and register your first AI agent. Each agent gets a unique ID that you pass in every SDK call.

💡 The Starter plan is free — up to 500 intents per month, no credit card required. Perfect for evaluating governance before going to production.
2
Install the SDK

Choose the SDK that matches your agent's runtime:

Terminal
# npm
npm install @aegis-os/sdk

# yarn
yarn add @aegis-os/sdk

# pnpm
pnpm add @aegis-os/sdk
Terminal
pip install aegis-os
3
Get your API key

In the dashboard go to Settings → API Keys and create a new key scoped to your agent. Copy it into an environment variable — never commit secrets to source control.

.env
AEGIS_API_KEY=aegis_sk_your_key_here
AEGIS_AGENT_ID=your-agent-uuid-here
4
Submit your first intent

Before your AI agent initiates any spend, submit an intent. The platform responds immediately — the policy engine evaluates asynchronously so your agent is never blocked.

agent.ts
import { AegisClient } from '@aegis-os/sdk';

const aegis = new AegisClient({ apiKey: process.env.AEGIS_API_KEY });

const intent = await aegis.intents.submit({
  agentId:    process.env.AEGIS_AGENT_ID,
  amount:     5000,
  currency:   'INR',
  reason:     'Purchase cloud storage subscription',
  confidence: 0.95,
  merchant:   'AWS',
});

console.log(intent.id);      // "int_9f3a2b..."
console.log(intent.status);  // "pending"
agent.py
from aegis_os import AegisClient
import os

with AegisClient(api_key=os.environ["AEGIS_API_KEY"]) as aegis:
    intent = aegis.intents.submit(
        agent_id   = os.environ["AEGIS_AGENT_ID"],
        amount     = 5000,
        currency   = "INR",
        reason     = "Purchase cloud storage subscription",
        confidence = 0.95,
        merchant   = "AWS",
    )
    print(intent.id, intent.status)
5
Handle the decision

The policy engine returns one of three outcomes. Act on each appropriately — only proceed with payment on approved.

✓ approved
● pending_approval
✕ denied
agent.ts
const resolved = await aegis.intents.waitForDecision(intent.id);

if (resolved.status === 'approved') {
  // Policy cleared this spend — safe to proceed
  await executePayment(resolved.id);

} else if (resolved.status === 'pending_approval') {
  // Finance team notified via dashboard — wait for their action
  console.log('Awaiting approval…');

} else {
  // Denied by policy — do not proceed
  console.error('Spend denied:', resolved.policyDecision);
}
💡 For approval flows that may take hours, configure a webhook in Settings → Webhooks instead of polling. Your endpoint receives a signed callback the moment a decision is made.
6
View the audit trail

Every intent and every state change is automatically appended to the immutable audit chain. Open the dashboard and go to Audit Logs to see the full history — timestamps, matched policy, risk score, and approver actions — all checksum-verified.

When you need to report to regulators, go to Export and choose your format: RBI, EU AI Act, SOX, or generic JSON / CSV. One click, ready to submit.

You're now governing AI spend end-to-end. Every rupee your agent touches is policy-evaluated, human-reviewable, and audit-ready.