Every dashboard action is an MCP tool your agent can call.Read the docs
Send & Retain

Let AI run your email. You approve the launches.

Your agent reads your contacts, automations, and delivery stats, then builds what's missing. Nothing reaches customers until you turn it on.

Start for free

✓Free to start. Unlimited contacts on every tier.

Works with

Copilot run
AI for email

Read, propose, then act

Operator prompt

Inspect the onboarding automation, explain where contacts stop, and show the evidence before proposing a change.

Read-only first task

Read

Live automation and evidence

Draft

Optimization proposed

Approval

Waiting for operator

Consequential writes need approval

[ 01 / 06 ]What it does

What can an AI agent do with your email?

The whole job, not just the copy. It can read everything, and every change is scoped and logged.

Get the domain sending

Create the domain, hand you the DNS records, verify them, and set up a sender.

Write and test templates

Draft react.email templates in your brand, render them, and send you a test.

Build email automations

Welcome, trial, and win-back emails triggered by what users do, created paused.

Wire up transactional email

Add receipts and password resets to your code with the email API.

Explain your numbers

Follow any email to inbox, bounce, or complaint, and compare automations week over week.

Move off your old platform

Read flows from Klaviyo, ActiveCampaign, or HubSpot and rebuild them as paused automations.

[ 02 / 06 ]Connect

How do I connect my AI agent?

One command in Claude Code. Every other agent uses the same URL and the same browser sign-in.

  1. Add the MCP server

    Run this in your terminal. No API key, no config file.

    Terminal
    claude mcp add --transport http sendandretain https://sendandretain.com/api/mcp

  2. Sign in through the browser

    Run /mcp, pick sendandretain, and approve. Your agent keeps the token, so nothing lands in a config file.

  3. Ask for a read-only audit

    Try: "Check my sending setup and tell me what's misconfigured." Then hand it a real change.

[ 03 / 06 ]Example prompts

What can you ask your agent to do?

Anything you'd ask a teammate who knows email. Start read-only.

  • Audit

    “Check my sending setup and tell me what's misconfigured. Don't change anything.”

  • Build an automation

    “Build a welcome automation for new signups in our brand. Leave it paused and show me each email.”

  • Template

    “Write a password reset email in our brand, render it, and send a test to me.”

  • Diagnose

    “Why did open rates drop last week? Break it down by automation and template.”

  • Weekly report

    “Summarise last week's sends, bounces, complaints, and clicks for each automation.”

  • Migrate

    “Read our Klaviyo flows and rebuild the welcome flow here as a paused automation.”

[ 04 / 06 ]Guardrails

Can I trust an AI agent with my email?

Yes, because the limits live in the product, not the prompt. Agents, apps, and people all send through the same checks.

  • Automations start paused. Turning one on needs your explicit confirmation.
  • No blasts. Every email goes to one person, so broad sending only comes from automations you turned on.
  • Complaint suppressions can never be removed. Not by the agent, not by you.
  • Every queued email re-checks the pause switch and daily cap right before it sends.
  • Don't want it making changes yet? Give it a "Read only" API key.
Delivery checks
Pre-dispatch

Refuse unsafe sends first

Details

not yet sent

To[email protected]

From[email protected]

Templatewelcome

Sourceapi

ProviderSend & Retain

Checks before dispatch

  1. DomainSPF · DKIM · DMARC
  2. RecipientAddress verified
  3. SuppressionNo active block
  4. RouteProvider healthy
Controls run again at dispatch

[ 05 / 06 ]Two ways in

Your own agent or the in-app copilot?

Both use the same tools and the same send checks. The difference is where they run.

Where it runs

In-app copilot

Inside the Send & Retain dashboard

Your agent over MCP

Claude Code, Claude, ChatGPT, Cursor, Codex, and other MCP clients

Which model

In-app copilot

Chosen by your tier

Your agent over MCP

Whatever your own plan runs

AI credits

In-app copilot

Metered on the tokens it uses

Your agent over MCP

None. It runs on your own plan

Best for

In-app copilot

Quick questions while you're in the dashboard

Your agent over MCP

Building automations, wiring events into your code, and weekly reviews

[ 06 / 06 ]Agents

Which AI agents work with Send & Retain?

Same account, same tools, same guardrails. Each page has the exact setup for that agent.

Questions people ask us

Hand the email backlog to your agent.

Connect it, ask for a read-only audit, then approve the first automation it builds.

Start for freeRead the MCP guide