Testing Product Launch Subject Lines in Zed
Zed's AI assistant can drive a full subject line A/B test for your next product launch when it's connected to marginal, the hosted email MCP server. Draft variants, launch the send, and read open/click lift — all from the editor you already code in.
Why run subject line tests from Zed
A product launch lives or dies on the open rate, and the subject line carries most of that weight. Instead of switching to a separate marketing tool, you can keep the loop inside Zed where you're already shipping the release. The AI agent reads your intent, calls marginal's tools, and hands back numbers you can act on.
Because marginal is a hosted MCP server, there's nothing to deploy. You point Zed at the endpoint, add your Bearer key, and the agent gains four tools for end-to-end testing.
- generate_variants — produce subject line candidates for the launch
- launch_test — start a managed A/B send across your list
- get_results — pull open and click metrics with within-test lift
- recommend_next — get a data-backed suggestion for the winner or a follow-up
Connecting marginal to Zed
Add marginal as an MCP server in Zed's assistant configuration. The endpoint is https://marginal.sh/mcp and authentication uses a Bearer API key in the marg_live_ format. Once registered, Zed's agent can call the tools directly during a chat.
The free tier covers 100 experiments per month, which is plenty for iterating on launch copy. Full setup details live in the docs at https://marginal.sh/docs/.
- Registry name: sh.marginal/mcp
- Endpoint: https://marginal.sh/mcp
- Auth: Bearer marg_live_... API key
- Reference: https://marginal.sh/mcp
A launch-day workflow
Treat the AI agent email marketing flow like any other task in Zed: describe the goal in plain language and let the tools do the work. A typical product launch sequence starts with variant generation and ends with a recommendation.
- Ask Zed to generate 4–6 product launch subject lines around your value prop
- Launch the subject line A/B test to a sample of your audience
- Wait for the send window, then call get_results for lift metrics
- Use recommend_next to roll the winning subject to the full list
Reading the results that matter
marginal reports open and click tracking with within-test lift, so you're comparing variants under the same conditions rather than guessing from absolute rates. For a launch, the open-rate lift on the subject line is the headline metric, while click lift tells you whether the framing also set the right expectation.
Keep the experiment focused: test one strong angle against a control rather than scattering many ideas. The agent can summarize the outcome in Zed and propose the next iteration, turning each launch into a faster, sharper subject line test.
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