Closed Loop Email Marketing AI
A closed loop email marketing AI doesn't just write copy — it sends, measures, and decides what to do next. marginal gives your AI agent the tools to close that loop over subject line A/B tests, all through a single email MCP server.
What 'closed loop' actually means for email
Most AI email tooling stops at generation: you ask a model for ten subject lines and copy them somewhere else. That's an open loop — the model never learns whether its ideas worked. A closed loop email marketing AI completes the cycle by feeding real send results back to the agent so the next decision is grounded in data.
marginal exposes that full cycle as MCP tools. Your agent proposes variants, launches a managed send, reads open and click metrics, and asks for a recommendation on what to ship next — without leaving the conversation.
- Generate candidate subject lines
- Launch a real subject line A/B test
- Track opens and clicks per variant
- Recommend the next action from measured lift
The four tools that close the loop
Each stage of the loop maps to one MCP tool. Because they share the same hosted backend, the agent works against consistent experiment state instead of stitching together disconnected services.
- generate_variants — draft subject line candidates for a campaign
- launch_test — start a managed send and split the audience
- get_results — pull open/click tracking and within-test lift metrics
- recommend_next — turn results into a concrete next step
Why this fits AI agent email marketing
Agents in Cursor, Claude Code, Windsurf, Cline, Continue, Zed, OpenAI Codex, and Claude Desktop can call these tools directly. That makes the experiment loop programmable: an agent can run a subject line A/B test on Monday, read the lift on Wednesday, and propose a follow-up variant — all as part of an ongoing growth task.
marginal is hosted, so there's nothing to self-host and no separate sending infrastructure to wire up. You authenticate with a Bearer API key and point your client at the endpoint.
- Endpoint: https://marginal.sh/mcp
- Auth via Bearer key (marg_live_...)
- Free tier: 100 experiments per month
- Works across every major MCP-capable client
Getting started with marginal
Connect your agent to the email MCP server at https://marginal.sh/mcp, then run a single experiment end to end to see the loop close. Start with one campaign, generate a handful of variants, launch the test, and let recommend_next guide your follow-up.
Full tool references and request examples live in the docs at https://marginal.sh/docs/ so you can wire up a repeatable closed loop email marketing AI workflow in minutes.
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