# Secure email gateways vs a pre-delivery inbox

> A gateway or filter in front of a mailbox, or an inbox that screens at its own door: where each fits when an AI agent reads the mail, and how to run both.

Source: https://trotta.email/compare/secure-email-gateways

Use a secure email gateway when you must protect many existing human mailboxes without moving them. Use an inbox that screens at its own door when an AI agent is the reader, because the inbox also decides what the agent can read, release and send. The two can run side by side.

Updated November 20, 2026

A secure email gateway is a filter that stands in front of mailboxes you already have. A pre-delivery inbox is a mailbox that does its own screening before it accepts a message. They solve overlapping problems from different places.

## Two kinds of filter in front of a mailbox

Filters that sit in front of a mailbox come in two broad designs. Ask a vendor which one you are buying.

- **In the mail path.** Your MX records point at the gateway. It screens each message and relays the rest to your mailbox provider. Screening happens before delivery.
- **Connected by API.** The filter reads mailboxes through the provider's API and removes mail it judges bad. Mail is usually delivered, then checked. Ask how long a message can sit in the inbox before it is pulled.

## A gateway and Trotta Mail side by side

|  | Gateway or filter in front of a mailbox | Trotta Mail |
| --- | --- | --- |
| Built for | Security teams protecting many human mailboxes. | AI agents and humans. The guard is stricter for an agent. |
| When mail is screened | In the mail path: before delivery. Connected by API: usually after delivery. Ask which. | While the sending server is still connected, before the message is accepted. |
| What an agent can read | Whatever the mailbox behind the gateway allows. Ask whether the gateway changes what an API key can read. | The safe view: hidden text removed, instructions aimed at a model marked, risky links defanged. |
| Held mail and human review | A quarantine, usually run by admins. Ask whether an agent's credentials could release from it. | Held outside the inbox. A human can release it. An agent's key cannot. |
| Outgoing checks | Often data loss rules, if outgoing mail is routed through the gateway. Ask. | Screened for secrets, sensitive data and replies steered to a new address. |
| Sign-in codes | Delivered to the mailbox like other mail. Ask who and what can read them there. | Sealed from every API key in a human's mailbox. Parsed for the agent in its own inbox. |
| Setup for many agents | A mailbox for each agent at your provider, plus the gateway's routing and policies. | One API call per inbox. Pods keep each customer's inboxes apart. |
| Price to start | Commonly per mailbox, on top of the mailbox itself. Ask for the minimum. | Free: 2 inboxes, 1,000 emails a month. Developer, $20 a month: 5 inboxes, 5,000 emails. Startup, $200 a month: 50 inboxes, 50,000 emails. |

## When a gateway is the better choice

- You have hundreds of human mailboxes at one provider and cannot move them.
- You need one policy, one quarantine and one report across the company.
- You need services around the mail path, such as archiving, encryption or data loss rules for each employee.
- Your readers are humans. Warning banners, link rewriting and awareness training are built for them.

## When an inbox that screens at its own door is the better choice

A gateway decides whether a message gets in. What an agent may do once the message is there is usually decided by the mailbox behind it. An inbox that screens at its own door decides both, which matters when the reader is a model.

- The agent is given a safe view, with hidden text removed and instructions aimed at a model marked.
- Held mail stays outside the inbox, and the agent's key cannot read or release it.
- Outgoing mail is screened for secrets and for replies steered to a new address.
- Sign-in codes in a human's mailbox are sealed from every API key.

No filter catches everything, at a gateway or at a door. A reworded scam with no technical tell can still be delivered, which is why uncertain mail should [wait for a human](https://trotta.email/learn/held-mail).

## Using a gateway and a pre-delivery inbox together

The two do not compete for the same mailboxes. Keep the gateway in front of your human mailboxes on your main domain. Put agent inboxes on a subdomain such as `agents.example.com`, whose MX record points at the inbox that screens at its own door.

Mail for the main domain stays where it is. Your security team keeps its policies, and each agent reads a mailbox built for it.

## What moving looks like

1. Leave the gateway and the human mailboxes alone.
2. Create an inbox for each agent with one API call.
3. On a paid plan, [connect the subdomain](https://trotta.email/docs/domains) by publishing five DNS records, then check them.
4. Change each agent's address, then remove its access to human mailboxes.

## Questions

### Is a pre-delivery inbox an alternative to a secure email gateway?

For agent mail, yes: the inbox does its own screening before it accepts a message. For a company's existing human mailboxes, a gateway is usually the practical choice. The two can run together.

### Do secure email gateways screen mail before delivery?

Gateways in the mail path do. Filters connected by API usually act on mail after it is delivered. Ask a vendor which design it sells, and how long a message can stay readable.

### Does a secure email gateway protect an AI agent from prompt injection?

Ask the vendor whether it looks for [text written to hijack a model](https://trotta.email/threats/prompt-injection-by-email), such as instructions hidden from a human reader. Then test it with a sample message.

### Do I need my own domain to use both?

No. Agent inboxes can start at trotta.email on the Free plan. Your own domain or subdomain needs a paid plan: up to 3 domains on Developer, up to 25 on Startup.

## Sources

- [Trotta Mail API summary for agents](https://api.trotta.email/llms.txt)
- Trotta Mail docs: concepts, security model, connect a domain, plans and limits

## Related

- [What is a secure email gateway?](https://trotta.email/learn/secure-email-gateway-vs-pre-delivery-inbox) Guides
- [Pre-delivery vs post-delivery screening](https://trotta.email/learn/pre-delivery-vs-post-delivery) Guides
- [Business email compromise](https://trotta.email/threats/business-email-compromise) Threat library
- [Pricing](https://trotta.email/pricing)
- [Security](https://trotta.email/security)
- [Quickstart for agents](https://trotta.email/docs) Docs

Every check is on, on every plan. Free for two inboxes. No card.

[Get an inbox](https://trotta.email/signup) [See pricing](https://trotta.email/pricing)
