Most marketing workflows need dependable automation more than an autonomous AI agent. If the steps are clear, mistakes are costly or someone will audit the process, automate it. Save agents for work with unpredictable inputs and real judgment calls. Then keep a person accountable for the decisions that shape your brand and your customers' trust.
Why is everyone suddenly selling agents?
Because the technology got genuinely better, fast. I'm not here to wave it away. I've used generative AI many ways by now, including synthesizing analytics from several platforms into reports leadership could actually read. It saved real time.
I've also been in marketing long enough to watch a few "this changes everything" waves arrive. Each one changed something. None of them changed the need for someone to decide what's worth doing
When is an automation the better choice?
Vika Guseva makes the case well in her Prompt to Pipeline essay, "Most Companies Don't Need AI Agents." Her rough rule of thumb:
A clear workflow points to automation.
A high cost of error points to automation.
A compliance review points to automation.
Only unpredictable inputs, branching decisions and actual judgment point toward an agent.
Her sharper observation is that many products sold as agents are automations with a language model attached and a bigger price tag.
She also notes that autonomy adds uncertainty, uncertainty adds review time, and review time adds cost. Anyone who has marketed in a regulated space will nod at that. I produced diagnostic content at BakoDx under strict healthcare compliance review, and every extra unknown in a process meant another round with reviewers.
How do you know what level your team is at?
One maturity model I saved describes six levels of AI use in marketing teams, from a one-off user typing single requests up to an orchestrator running several agents in parallel. Its claim is that most marketing leaders sit at level two: brand voice and instructions are set up, but every task still starts manually.
Level two is a fine place to be. The next step up is usually a process change, turning repeated prompts into documented workflows, well before it's a new tool. Climb when the work calls for it.
How should you evaluate a new AI marketing tool?
Ask the vendor to describe, step by step, what the tool does with your data and where a person can review it. If that explanation gets fuzzy, slow down. One LinkedIn post I saved, mapping legacy GTM tools to modern replacements, ends with a caveat I respect: whether a newer tool should replace an older one depends on the workflow, the data, the budget and how much automation you actually need.
FAQ
What's the difference between an AI agent and an automation?
An automation follows steps you define, the same way every time. An agent decides its own steps toward a goal you give it, which makes it flexible and harder to predict.
Is it safe to let AI send emails to customers?
For confirmations and routine updates, with templates you've approved, often yes. For anything that represents your judgment or your brand voice in a new situation, keep a person reviewing it.
Where should a small team start with AI?
Pick one repetitive task that eats hours every week, document how it's done today, and automate that. Measure the time saved before adding anything else.
Sources
Vika Guseva, "Most Companies Don't Need AI Agents," Prompt to Pipeline https://lnkd.in/p/eqTXX9DY
LinkedIn post describing six levels of AI and Claude Code maturity in marketing teams
LinkedIn post mapping legacy GTM tools to modern replacements
Thinking through an AI decision?
If you're weighing an agent, an automation or simply leaving a process alone, I'm glad to think it through with you. I've been on the hopeful side and the skeptical side of these tools, sometimes in the same day. Connect with me on LinkedIn or read more about my background on my About page. Let's talk.