Give your AI agent
Identity infrastructure for AI agents.
One API, every capability.
Today, agents operate with borrowed identities and hardcoded secrets.
No email
Agents use your Gmail or dump messages to a queue. No real address, no threading, no DKIM.
No credentials
API keys in .env files. No rotation, no 2FA, no vault. One leaked key breaks everything.
No accountability
No delegation chain. No way to trace an agent's action back to the human who authorized it.
Agents are the new users.
They need identity.
Humans need an email address, a password manager, and a phone for 2FA to do anything meaningful on the internet. AI agents need the same primitives — plus delegation chains and audit trails that trace every action back to the human who authorized it.
Why agents need identityHuman
Agent (Mailgent)
One identity. Six capabilities.
Each identity is a hub — attach the capabilities your agent needs.
Voice · soon
Phone calls & SMS
Pay · soon
Payments & wallets
Social · soon
Social accounts
Messaging · soon
WhatsApp, Telegram
What agents build with Mailgent.
Every use case combines email, vault, and 2FA — the full identity stack.
Customer Onboarding
Agent signs up on platforms, verifies email, stores credentials, and confirms setup — fully autonomous.
Meeting Scheduling
Agent reads incoming requests, checks availability, sends calendar invites, and follows up with confirmations.
Document Processing
Agent receives documents via email, logs into processing platforms, uploads files, and sends results back.
Payment Link Delivery
Agent generates Stripe payment links using stored API keys and emails them directly to clients.
Compliance Monitoring
Agent tracks regulatory deadlines, logs into portals to check status, and alerts before anything lapses.
Sales Follow-ups
Agent sends timely, personalized follow-ups from its own inbox so warm leads never slip through the cracks.
AI agents in regulated industries
need provable identity.
FINRA fined 16 firms $1.1B for communication attribution failures. HIPAA mandates unique identification for every system touching patient data. CMMC blocks DoD contracts without it. The pattern is the same everywhere: if an AI agent acts, someone needs to prove who authorized it.
Finance
FINRA · SEC · SOX
Healthcare
HIPAA · Part 11
Legal
FRCP · FRE 502
Defense
CMMC · NIST 800-171
Insurance
UCSPA · RESPA
Three steps. Two minutes.
Click each step to see the code.
const API = "https://api.mailgent.dev/v0"
// Create an agent identity via the console API
const res = await fetch(`${API}/identities`, {
method: "POST",
headers: {
"Authorization": `Bearer ${process.env.MAILGENT_API_KEY}`,
"Content-Type": "application/json",
},
body: JSON.stringify({
name: "Sales Agent",
emailName: "sales",
scopes: ["mail:read", "mail:send", "vault:read"],
}),
})
const { identityId, emailAddress, rawKey } = await res.json()
// identityId: "id-f3875dadcb174bb2"
// emailAddress: "sales-x8k2m@mailgent.dev"
// rawKey: "mgent-..." (shown once, save it!)Native MCP server.
Connect any AI client.
Mailgent ships a built-in MCP server with mail.* namespaced tools. Connect Claude Desktop, Cursor, Claude Code, or any MCP-compatible client directly to your agent's capabilities.
{
"mcpServers": {
"mailgent": {
"url": "https://api.mailgent.dev/mcp",
"headers": {
"MAILGENT_API_KEY": "mgent-your-api-key"
}
}
}
}Every action traces back
to the human who authorized it.
Human > Org > Agent > Action. Cryptographically verifiable via DKIM. Revoke the human and every agent stops. No orphaned access. No cleanup scripts.
Human (ops@novaworks.io)
└─ Org (novaworks)
└─ Identity (sales-agent-x8k2m)
└─ Action (email sent, credential accessed, TOTP generated)
Every outbound email carries this chain in DKIM headers.
Every credential access logs it in the audit trail.
Revoke the human? → Org loses authorization.
Revoke the org? → All identities shut off.
Revoke the identity? → All capabilities stop.DKIM-signed identity
Every outbound email is DKIM-signed with agent identity headers. Recipients and regulators can cryptographically verify who sent it.
Instant revocation
Revoke a human, org, or identity — everything downstream stops instantly. No dangling credentials, no orphaned access.
Built on 40-year-old standards
DKIM, SPF, DMARC, OAuth 2.1, MCP. No proprietary protocol. Email's global PKI is already deployed everywhere.
Compliance-ready audit trails
Every email, credential access, and TOTP code is logged with identity, timestamp, and delegation context. The trail regulators ask for.
Works with every framework.
REST API + native MCP server. Framework-agnostic by design.
30 seconds to
first email.
Create an identity. Send your first email. No credit card.
# 1. Sign up at console.mailgent.dev and get your API key
# 2. Create your first identity
curl -X POST https://api.mailgent.dev/v0/identities \
-H "Authorization: Bearer mgent-YOUR-KEY" \
-d '{"name": "My Agent", "emailName": "agent"}'
# 3. Send your first email
curl -X POST https://api.mailgent.dev/v0/inboxes/agent-xxxxx@mailgent.dev/messages/send \
-H "Authorization: Bearer mgent-YOUR-KEY" \
-d '{"to": ["hello@world.com"], "subject": "Hello!", "text": "It works!"}'
# ✓ Done. Your agent has an identity.