AI Agent Email Marketing: Run Tests From Your Assistant
AI agent email marketing means giving your coding assistant or autonomous agent the ability to write subject lines, launch experiments, and act on real open and click data — all through a single tool interface. marginal makes that practical with a hosted email MCP server your agent can call directly.
What an AI agent can do for email
Most email tooling assumes a human in a dashboard. AI agent email marketing flips that: the agent reads campaign context, drafts variants, and runs the test loop without you copying text between tabs. marginal exposes the exact operations an agent needs as MCP tools, so the model can reason about results instead of guessing.
- generate_variants — draft multiple subject lines from a prompt or brief
- launch_test — start a subject line A/B test with a managed send
- get_results — pull open/click tracking and within-test lift metrics
- recommend_next — let the agent suggest a follow-up variant or winner
Why a hosted email MCP server matters
An agent is only as useful as the tools it can reach. marginal is a hosted email MCP server at https://marginal.sh/mcp, so there is nothing to deploy or self-host. Your agent authenticates with a Bearer API key (marg_live_...) and immediately has access to sends, tracking, and experiment results.
Because the send and measurement layer lives in marginal, the agent works with structured outcomes — variant-level opens, clicks, and lift — rather than scraping reports. That keeps the agent's decisions grounded in the same numbers you'd check yourself.
A subject line A/B test, agent-driven
Here is the loop your agent runs. Ask it to draft options for a campaign, and it calls generate_variants. It launches a subject line A/B test with launch_test, then waits for engagement to accumulate. When you ask how the test is doing, it reads get_results and reports the leading variant by lift, and can call recommend_next to propose what to send to the remaining audience.
- Describe the campaign and audience in plain language
- Approve or edit the variants the agent proposes
- Let the agent launch and monitor the test
- Act on lift metrics instead of subjective preference
Getting started with marginal
marginal works with the agents you already use — Cursor, Claude Desktop, Claude Code, OpenAI Codex, Windsurf, Cline, Continue, and Zed — through standard MCP configuration. The free tier covers 100 experiments per month, which is enough to validate AI agent email marketing on real campaigns before scaling up.
Point your client at the marginal endpoint, add your API key, and your agent gains email tools instantly. See the docs at https://marginal.sh/docs/ for client-specific setup and tool reference.
Get started with marginal
marginal is a hosted email marketing MCP server at marginal.sh. Sign up free, create an API key, and connect https://marginal.sh/mcp from Cursor, Codex, or Claude Desktop.
- 100 experiments/month on the free tier
- Four MCP tools: generate_variants, launch_test, get_results, recommend_next
- Listed in the MCP Registry as sh.marginal/mcp