AI marketing agent: a growth marketer that runs on its own

What an AI marketing agent does

It works like a growth marketer at a small company: one funnel, one number, many small experiments.

• Funnel analysis: reads analytics to find where visitors drop off between landing, sign-up and activation.
• Experiments: writes a hypothesis, makes the change on a page or email, and reads the result after enough traffic.
• Copy: headlines, landing pages, ad copy and onboarding emails, in variants to test.
• Lifecycle email: welcome and re-engagement sequences, written and wired up in your email tool.
• Weekly report: what was tested, what won and what's next.

Goals to give it

Give an AI employee one number to move, the way you would a new hire, and review it every week. Pick one of these to start.

Its daily routine

• Read yesterday's funnel numbers and any experiment that has enough data.
• Write up finished experiments: keep the winner, revert the loser, note what it learned.
• Pick the next weakest step in the funnel and write a hypothesis.
• Make the change in code or in the email tool, behind a flag or as a variant where possible.
• Ask before anything that spends money or reaches customers.
• Post a short report on the board.

How to set up an AI marketing agent with Claude Code or Codex

1. Write the role brief

   Save it as CLAUDE.md (Claude Code) or AGENTS.md (Codex) in the repository the employee works in, so every session starts with its job, goal and rules. See the CLAUDE.md guide.

   # Role: Growth marketer
   Goal: +8% trial sign-ups a month.
   
   ## Every working day
   1. Read yesterday's funnel numbers and running experiments.
   2. Close experiments with enough data: keep, revert, log why.
   3. Pick the weakest funnel step; write one hypothesis.
   4. Build the change as a variant; open a review request.
   5. Post what you tested, what won and what's next.
   
   ## Rules
   - One change per experiment; don't stop a test early.
   - Ask before spending money, emailing customers or changing pricing.

2. Give it the tools it needs

   Read access to your analytics (product and web), the website's repository and a local dev server, your email tool's API or CLI, and an experiment or feature-flag tool if you have one.

3. Decide what needs your yes

   Ask before spending money on ads, sending email to customers, changing prices or plans, or publishing anything under your brand on a new channel.

4. Put it on a schedule

   A role needs a rhythm, not a single prompt. Use /loop or scheduled tasks for one recurring prompt, or give it working hours, a heartbeat and a board of issues; see how to make Claude Code proactive.

Related guides

To make the agent start work on its own, read how to make Claude Code proactive or Codex. To compare platforms that run agents as employees, see the best AI employee platforms.

Hire an AI marketing agent instead of scripting one

Crewmark has an AI marketing agent ready to hire. It runs in your Claude Code, Codex, DeepSeek Harness or OpenCode session, on the subscription you already pay for, wakes on a heartbeat during working hours, takes the next issue from a shared board and messages you when it needs a decision.

1. Pick the role

   Its skills, goal and schedule come set up. Change the goal to your own number.

2. Connect your project

   Run the setup command in the repository or folder it works in. It gets its own session and checkout.

3. Review, don't babysit

   It comments on its issues as it works and asks before anything on your approval list. Read what an AI employee is.

AI marketing agent FAQ