Markigy: The Science of Marketing Strategy · Leanne Dow-Weimer

Using AI agents to to reshape MOPS & More w/ Stephen Stouffer

·34 min·2 clips

Managers may soon oversee people, consultants, and AI agents at the same time.

1. Markigy: The Science of Marketing Strategy focuses this episode on using AI agents to reshape MOPS and marketing work. 2. Leanne hosts Stephen Stouffer, who describes himself as working in ops at his company and building agents across sales, marketing, and the website team. 3. The episode asks what AI agents change when teams stop asking “how do we get to the end state” and start defining the end state instead. 4. Stephen says ChatGPT has been “all-encompassing” over the last couple of years since OpenAI brought it to the world. 5. He describes a shift from using agents for content generation and email tone changes to having them write code for HTML and CSS. 6. He gives a concrete workflow where agents help upload records to HubSpot or Salesforce and create UTM-enriched campaign URLs. 7. Leanne compares that with earlier Zapier-style automations and asks how robotic process automation differs from an AI agent. 8. Stephen separates automation into logic-based automation, AI-infused automation, and full-on AI agents. 9. He says the key difference is that an agent is not following a prescriptive if-else path and instead decides which tools to use to complete the job. 10. He says guardrails matter because some models return a score from 0.0 to 1 showing how well a result matches the request. 11. He also says that access controls matter, because an agent with internet search can be pulled toward random results from Google or even a Reddit thread. 12. Stephen recommends a trusted vector database, Notion, or Google Docs as a safer source of knowledge for agent responses. 13. Leanne pushes on hallucinations and contextual errors, including the example of giving different GDPR guidance for North America versus the EU. 14. Stephen says AI can give a result “very confidently” without asking clarifying questions, which makes human review important. 15. He recommends using AI in narrowly scoped ways, such as feeding it the last 10 email sends, subject lines, HTML, and performance metrics. 16. He warns that bad data or fake data will lead to bad decisions, even when the AI output looks polished. 17. The tone stays conversational and skeptical, with Leanne repeatedly testing Stephen’s claims and Stephen answering in practical terms. 18. The format stays interview-driven and tactical, with examples from meetings, Slack channels, customer success flags, and lead routing. 19. Marketing operators, rev ops teams, and managers trying to add AI to workflows would likely get the most out of it. 20. People looking for abstract futurism without day-to-day workflow examples may skip it.

Listen to the show on

Apple Podcasts Spotify

Play clips from this episode More from Markigy: The Science of Marketing Strategy