Sam Brovda comes from enterprise sales and builds software on the side using AI coding agents. After launching a consumer mobile app, he ran into a familiar problem: the product worked, but nobody heard about it. Without a marketing team or budget, he decided to build one using AI bots. In this article, he explains how he runs marketing with Grok Bot without writing code, using specialized bots, written guardrails, and a strict rule that nothing publishes without his approval.

What Happened

Instead of relying on one general assistant, Brovda created several bots, each responsible for a single job. The marketing bot generates content ideas every weekday morning, ranks them, and drafts posts when given the go-ahead. A chief of staff bot reviews the actual output of the other bots before 8 a.m. each weekday and flags anything that is slipping or stalled. A separate bot writes instructions for new bots, making sure each one has a defined voice and a clear list of things it must never do.

The marketing bot never posts, comments, or messages anyone. It only drafts. Brovda reviews the draft and publishes it himself. He compares the arrangement to sales: a representative can draft a proposal, but only leadership signs the contract.

Every job reads a shared rules file before doing anything. The guardrails prohibit spending on ads or paid creators, ban invented quotes or statistics, and require that every product claim be true today. A weekly Friday report adds dated lessons, such as logging posts the same day they go up and checking engagement about an hour after publishing. Because there is no signup page yet, calls to action shifted from signups to follows or comments. The system also keeps a 21-day content bank so he never starts from a blank page.

Why This Matters

Hiring a marketing team is often unrealistic for solo builders and small startups. This approach shows how AI agents can handle planning and drafting while a human keeps final authority. Separating drafting from publishing limits reputational risk. The rules file serves as institutional memory, preventing repeated mistakes across bot runs. It also proves that AI marketing does not require paid tools or a large budget—only clear instructions, firm boundaries, and consistent oversight.

Key Takeaways

  • Give every bot one specific job and one consistent voice.
  • Maintain a shared rules file with hard guardrails and dated lessons.
  • Never allow an AI bot to publish under your name without human approval.
  • Review actual output rather than trusting status reports.
  • Start with several narrow bots instead of one general assistant.

Conclusion

Brovda's setup offers a practical model for running marketing with AI agents and no code. The formula is straightforward: narrow bot roles, written boundaries, a human in the loop, and a system that learns from its mistakes. Even this article followed that process—a bot drafted it from the work described, and it was published only after human approval.