AI Services Market › Use cases › Sales, Marketing & Support
Sales, Marketing & SupportDone with Claude Code Oct 17, 2026
Run Meta ad campaigns autonomously for a newsletter
A Claude Code agent generated ad images, managed Meta campaigns via API, built landing page variants and pulled analytics, testing about 50 variants over 31 days.
- What They Did
- The author gave a Claude Code agent $1,500 and a Meta Ads account for a newsletter. Each morning he typed one command. The agent read its own prior logs, pulled Meta and PostHog data, decided whether to create, pause or scale ads, executed, and logged its reasoning to git. He intervened once, adding email validation to the lead form.
- How It Went
- It spent $1,493 and got 243 leads at $6.14 each, missing the $2.50 target. Whiteboard-style ads won. The one human change spiked cost per lead above $50, and the account never fully recovered.
- Worth Knowing
- Telling the agent the experiment ended at day 30 made it play safe, and it ignored lead quality until day 16, so frame the goal as ongoing and specify quality upfront.
Try it yourself with Claude Code
Help me run a small Meta ads test for [my newsletter or product]. Generate [number] ad image and copy variants aimed at [audience], build matching landing page variants, and set up campaigns as paused drafts through the Meta API. Pull results into a simple report ranked by [metric, e.g. cost per signup]. Show me every campaign, budget, and audience before anything is activated or any money is spent.
Do this with a skill.
These AI Services Market playbooks do this exact job.