Private Beta — Limited spots available

Stop your AI agent
from going rogue.

Auditail wraps every action your AI agents take with guardrail rules and records each decision in a tamper-evident audit trail: the evidence your SOC2 auditor asks for. Drop-in SDK for JavaScript and Python.

TypeScript

import { Auditail } from 'auditail-sdk';

const aegis = new Auditail({ apiKey: 'your_key' });

aegis.addRule(Auditail.rules.noExternalEmail());

const result = await aegis.guard({

action: 'send_email',

payload: { to: '[email protected]' },

agentId: 'financial-bot-01',

});

// result.approved → false

// result.reason → "Blocked: external email is not permitted"

// result.auditId → "aeg_1a2b3c_x9z4k7" ← logged forever

🚀 Request Early Access

Try it

Be the agent. See what Auditail does.

Pick an action a payments agent wants to take. Rules decide instantly, risky ones wait for a human, and every decision lands in a log that cannot be quietly edited.

1 · Agent actions simulation

Rules in force

payout-over-limit pauses payouts above $10,000 until a human approves
no-external-email blocks email outside @acme.com
no-data-deletion blocks DELETE, DROP, TRUNCATE

// rule payout-over-limit: action 'require_approval' const result = await auditail.guard({ action: 'create_payout', payload: { amount: 25000 }, agentId: 'payout-agent', }); if (!result.approved) return result.reason;

2 · Audit log hash-chained

No events yet. Pick an action on the left.

Run an action that gets blocked, then try to hide it.

This demo runs entirely in your browser and uses the same hash-chaining scheme as the real service. In production, events are stored on our servers, shown in the dashboard, and checked with one click.

Why Auditail

Your AI is ready for production.
Is your legal team?

Developers currently build custom logging, permission checks, and audit systems from scratch — wasting months before shipping. Auditail is the layer they all end up building.

🛡️

Real-time Guardrails

Block dangerous actions before they execute. No external emails, no DB deletions, no PII leaks — enforced at the SDK level.

📋

Tamper-evident Audit Trail

Every action, every decision, every payload — stored in a hash-chained log. Edit or delete a past event and the integrity check flags exactly where. Reference the audit ID in any incident review.

🔔

Live Dashboard

Watch every allowed and blocked action as it happens, filter by agent or rule, and drill into the exact payload. Check the integrity of the whole log with one click.

👤

Human Approval

A rule can pause the agent until a person approves or denies the action. Approve in the dashboard or get pinged in Slack, Teams or Discord through a signed webhook. No answer means denied, and the decision is written to the tamper-evident log.

📄

SOC2 Audit Evidence

Auditors want proof of what your systems did and what was allowed. Auditail records every agent action, the rule that decided it, and the full payload in a log you can verify at any time. Export any period and your auditor can verify it offline with a free tool, no Auditail account needed. It supports your audit; it does not replace your controls or certify you.

🔏

Signed Receipts

Every human approval comes with a signed receipt. Your payment service checks it before moving money, so an agent can't skip the approval step, and a swapped amount or beneficiary is rejected.

🏠

Private Cloud roadmap

Run the data plane inside your own cloud account so event data never leaves your infrastructure. Planned for design partners who need it.

How it works

One method. Full protection.

Wrap every AI agent action with aegis.guard(). That's it.

Step 1

AI decides to act

send_email, run_query, http_request…

→

Step 2

aegis.guard()

Rules evaluated in <5ms

→

Step 3

Decision + Log

Hash-chained audit event stored

✅ Approved → Agent executes

🚫 Blocked → Agent aborts + team alerted

Pricing

Usage-based. Scales with you.

Planned pricing, usage-based with no per-seat fees. Design partners get early access and lock in their rate.

Hobby

Free

For solo developers

✓ Up to 10,000 logged actions / mo
✓ Core SDK (JS + Python)
✓ Community support

Request Early Access

We're onboarding design partners now. Answer 3 quick questions and we'll be in touch within 24 hours.

Work email *

What AI framework are you currently using? *

What's your biggest fear about putting your AI agent in production? *

When do you plan to deploy to production? *

No spam. We'll reach out personally via email within 24 hours. During the private beta, all keys are free.

✅

You're on the list!

We'll review your answers and reach out within 24 hours with your API key and onboarding instructions.

In the meantime, star the repo on GitHub and try the preview SDK — it works fully without a key.

Stop your AI agent from going rogue.