DTC Email Experimentation, Driven by Your AI Agent
Direct-to-consumer brands live and die by inbox performance. marginal lets you run DTC email experimentation directly from your AI agent — generating subject line variants, launching tests, and reading lift without leaving your editor.
Why DTC brands need faster experimentation
DTC email programs send a lot: launches, restocks, win-back flows, and promo blasts. Each send is a chance to learn, but most teams skip testing because the tooling lives in a separate dashboard and slows the workflow down.
marginal removes that friction by exposing experimentation as an email MCP server. Your AI agent can propose a subject line A/B test, ship it, and surface results — turning every campaign into a small, measurable experiment.
- Test promo angles (discount vs. free shipping) before a full send
- Compare urgency framing for restock and back-in-stock emails
- Iterate on win-back subject lines using real open and click data
- Build a habit of testing without context-switching to another app
How marginal fits an AI agent workflow
marginal is a hosted MCP server reachable at https://marginal.sh/mcp. Once your agent is connected with a Bearer API key, AI agent email marketing becomes conversational: ask for variants, launch a test, then ask which one won.
The MCP tools map cleanly to the experimentation loop you already run by hand.
- generate_variants — draft subject line options for a campaign
- 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 — get a data-backed suggestion for the next iteration
A practical DTC testing loop
A typical DTC email experimentation cycle in marginal starts with a campaign brief and ends with a recommendation you can apply to the next send. Because everything happens through your agent, the loop is fast enough to run on every meaningful campaign.
Connect one of the supported clients — Cursor, Claude Desktop, Claude Code, Windsurf, Cline, Continue, Zed, or OpenAI Codex — and the same tools are available wherever you work.
- Describe the email and audience to generate candidate subject lines
- Launch a subject line A/B test as a managed send to your list
- Wait for opens and clicks to accumulate, then read lift
- Apply recommend_next to refine the angle for your following campaign
Start small and scale your tests
The free tier includes 100 experiments per month, which is enough to test every major DTC campaign and build a real picture of what your audience responds to. There's nothing to self-host — marginal is fully managed.
See the docs at https://marginal.sh/docs/ for tool schemas and connection details, then point your agent at https://marginal.sh/mcp to begin.
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