How to Identify the Right AI Use Cases for Your Marketing Team

Struggling to find your next AI use case? You may just be looking in the wrong place.

The best opportunities for AI often aren’t net-new things your team could be doing, but things your team already does over and over again.

These recurring tasks often include researching markets, evaluating creative, analyzing performance, drafting communications, synthesizing customer feedback, or preparing reports.

That was one of the central ideas in our webinar, The Skill to Build Skills, led by Jack Atlasov, Director of Agentic Commerce at Right Side Up. 

Jack explored a simple question that’s not so simple to answer: How can marketers turn one-off AI tasks into repeatable wins?

Let’s take a look at the webinar’s most important takeaways, including the steps to identify your next AI use case, how to give AI the right context, and real-life examples of what the process can look like.

[tldr]

TL;DR: Start with repeated work, not hypothetical AI use cases

Look for work that is:

• Repeated frequently: The same or similar task happens again and again.
• Judgment-heavy: Doing it well requires context or expertise that lives in your team’s heads but isn’t always thoroughly documented.
• Passed between people or sessions: This context needs to survive a handoff (or several handoffs).
• Measurable: You can define what a better outcome would look like.

Those are strong candidates for reusable AI workflows or “skills”: systems that preserve instructions, context, resources, and sometimes tools so the work doesn’t have to start from zero every time.

[/tldr]

Why repeated work is the best place to look for AI leverage

Most marketing teams have some degree of AI proficiency.

They can give a model a content brief and get usable copy or feed it well-formatted research to obtain a helpful analysis.

The harder question is whether the next person (or even the same person the next time they’re in Claude) can reproduce a high-quality result without rebuilding part (or all) of the context.

This challenge is why AI in marketing is starting to shift more towards building operating systems, and why teams need to look for AI opportunities differently.

"AI in marketing is moving from answers to operating systems." —Jack Atlasov, Director of Agentic Commerce, Right Side Up

Instead of starting with a blank whiteboard and brainstorming everything AI could theoretically do, start by observing what your team already asks it to do.

Jack offers a simple rule of thumb: keep a running list of tasks you hand to AI more than three times in a month. That list becomes the beginning of a “skill backlog.” The more frequently a task appears, the stronger the signal that it may be worth systematizing.

How to identify a high-value AI use case

Jack outlines three key factors that help identify the most compelling AI opportunities.

Frequency is the starting point, but it’s not the only criteria that dictates whether a new workflow is warranted or not.

1. Is the work repeated often?

A task that happens once a year probably doesn’t need a sophisticated reusable system. Use the “3x per month” rule to guide you, and think beyond obvious production tasks, too. 

Repeated work might include reviewing briefs, preparing performance summaries, analyzing competitors, assessing whether content meets brand standards, or synthesizing information before a decision.

A simple question to answer here: Where does your team keep solving essentially the same problem?

2. Does the task require meaningful judgment?

This is where the opportunity can get more (or less) interesting.

Some of the most important marketing processes hold their value not from the number of steps required, but from all the context that has to be applied during execution—context that isn’t always written down by experienced marketers. This can include things like: 

• What “on-brand” really means
• Which performance change is worth escalating
• Which customer insight is truly meaningful 
• What makes a strategy presentation credible to the executive team

Jack described this as implicit or tacit knowledge: the judgment sitting in people’s heads that a general-purpose AI model doesn’t automatically have.

A strong AI workflow captures and encodes that so the system has a better chance of applying the same standards consistently.

3. Does the work cross people, tools, or sessions?

A marketer might develop a highly effective way of prompting an AI tool, but if the workflow lives entirely in their chat history, the organization hasn’t really captured the value.

The same problem appears when someone has to re-explain context every time they open a new AI session, or when work passes from one team member to another and the reasoning behind previous decisions disappears.

When repeated work depends on context surviving across people, platforms, or time, a reusable system becomes much more valuable.

The hardest part is giving AI the right context

Once you identify the workflow, the next step is making it usable by AI—and it’s more challenging than you may think.

“Most teams aren’t really blocked by prompting or accessing AI. They’re usually blocked by context.”—Jack Atlasov, Director of Agentic Commerce, Right Side Up

A reusable workflow needs to capture a lot of what an experienced marketer already knows.

That might include examples of:

• Strong and weak outputs
• Definitions your team uses differently from the rest of the industry
• Brand rules
• Approval criteria
• Source material
• Constraints
• Tools
• Escalation logic
• The standards someone uses to decide whether the result is actually good

Jack suggested thinking about the process more like briefing a new employee. 

You wouldn’t hire someone, give them a single sentence of instruction, and expect them to immediately reproduce the judgment of your best marketer. AI is no different.

What this looks like in practice: from repeated task to reusable skill

During the webinar, Jack demonstrated the process by building a personal-voice skill. The goal: to translate fuzzy human judgment (“these materials sound like me”) into enough explicit context that the system could reproduce it more consistently.

He started with evidence, providing examples of his communication across email, Slack, and LinkedIn. 

He then gave the AI context about what a good skill should look like and used those examples to identify patterns, rules, and characteristics of his voice.

Then came an important step: evaluation. The workflow generated a scoring system so Jack could compare outputs, provide feedback, and determine whether the skill actually performed better than the baseline model.

In another example, he started with an existing open-source trend research skill rather than building from scratch. He had the AI inspect the existing system, plan ways to make it more useful for growth marketers, customize it, test it, and package the result for reuse.

The pattern was the same both times:

Identify the job → provide context → encode the method → test the output → improve the system.

[tldr]

See the skill-building process in action 

Watch the full webinar, The Skill to Build Skills, to see Jack build, customize, and evaluate reusable AI skills step by step.

[/tldr]

Don’t automate until you can define what “better” means

This was alluded to in the sample workflows Jack built, but it bears repeating: Don’t assume an AI workflow is valuable simply because it works.

“Don’t just assume that a skill is gonna work and create value just because you made it.”—Jack Atlasov, Director of Agentic Commerce, Right Side Up

Before automating or scaling a workflow, define the baseline:

• What happens today? 
• How long does it take? 
• What does a good output look like? 
• Which errors matter most? 
• What judgment should the system preserve?

In the end, the big question you’re trying to answer is: Did we materially improve the work?

This is especially important for client-facing, high-value, or high-judgment workflows. 

[tldr]

The best AI strategy may start with an audit, not a brainstorm

For marketing leaders wondering where to invest next in AI, the answer may already be visible in the way your team works today. Tasks that meet these criteria often warrant an AI workflow, or at least a discussion about it:

• We do this work repeatedly.
• Expert judgment plays an important role in the process.
• Critical knowledge currently lives inside individual marketers’ heads.
• Handoffs often force us to reconstruct context.
• We can define what would make an AI-assisted version meaningfully better.

Watch the full webinar for a hands-on look at the process. And if you need a helping hand, Right Side Up can help identify the highest-leverage opportunities, prioritize what’s worth building, and implement AI systems your marketing team can actually use. Get in touch today →

[/tldr]

‍