AI Operations: Using AI to Scale Your Work

Are you looking to streamline your business processes with AI? Want to learn how to create AI playbooks that can transform your operations to be more efficient?

In this article, we explore how to use AI to scale your work.

Why AI Operations Matter for Marketers

AI (artificial intelligence) operations, or AI ops, refers to the practice of using AI tools to execute work and improve business processes. While many marketers have begun using AI for content creation, AI ops take this concept further by applying AI across your workflow. This comprehensive approach can lead to significant improvements in efficiency and output quality.

One key mindset shift needed to leverage AI effectively is understanding that AI isn't just helpful for tasks you're not good at. It can be even more powerful when used to scale what you and your team excel at—your core competencies. This perspective opens up new possibilities for integrating AI into marketing operations.

How to Use AI to Scale Your Marketing Operations

#1: Getting Started with AI Operations

To start using AI operations for marketing, identify quick wins by looking for repetitive tasks you spend at least four hours per week on. These are good candidates for AI automation because they're frequent enough to justify the time investment in setting up an AI process but not so complex that they require constant human oversight.

Next, imagine what you could accomplish if time were no object. AI may allow you to execute ideas that weren't possible before due to time constraints. Ask yourself: “If I had unlimited time, what would I do for this client, project, or campaign?” Consider what data you'd analyze, research you'd conduct, or strategies you'd explore in-depth.

Instead of just using AI to fill gaps in your skill set, think about how you can use it to amplify and scale your proven methods and processes. For example, when auditing a client's social media strategy, you might typically only have time to review 10–20 posts. With AI, you could analyze the last thousand posts and apply your scoring method to identify high-ranking topics, resonating content, and off-brand material. 

These AI-powered “unlimited time offers” have numerous applications. You might use them in your sales process, client onboarding, or even in outreach to potential clients to showcase your capabilities. 

Documenting your processes is a crucial step in implementing AI operations. Create detailed playbooks that outline your methods step-by-step. These playbooks serve as the foundation for training AI on your specific workflows. They should include not just the actions to be taken but also the reasoning behind them and the criteria for success. 

For example, AI strategist Rachel Woods created an experimental playbook focused on LinkedIn engagement. 

The challenge she and her team tackled was crafting authentic, non-intrusive comments on LinkedIn posts. As many professionals have experienced, LinkedIn can be plagued by generic, AI-generated comments that add little value to discussions. Rachel and her team wondered if they could use AI to generate more meaningful engagement while still maintaining a personal touch.

They created a playbook designed to mirror Rachel's thought process when responding to comments on her LinkedIn posts. The goal was to see if AI could capture the nuances of her engagement style, including providing additional value, appreciating others' contributions, and asking thoughtful follow-up questions.

This playbook included detailed guidelines on:

• Analyzing the context of the original post
• Understanding the intent and tone of the comment
• Selecting an appropriate response technique
• Drafting a personalized reply
• Ensuring the response aligned with Rachel's voice and brand

The AI-generated responses were “scarily good,” with about 70–75% of the comments requiring no editing before posting. 

Note: You don't need to overhaul everything at once. Start small by focusing on one process and expand from there.

The key to success is thinking strategically about how AI can impact your marketing operations. Focus on time-consuming tasks that don't require high-level strategy or creativity. These might include data analysis, initial content drafts, or routine client reporting.

#2: Choose Which Tasks to Delegate to AI

When deciding which tasks to delegate to AI, use this three-level hierarchy that helps clarify which jobs are best suited for AI assistance:

Objective Tasks: These have clear right or wrong answers and are the easiest for AI to handle with proper instructions and examples. Examples might include categorizing social media posts by topic or calculating engagement rates for content.

Good-Enough Tasks: These require meeting certain criteria but have some flexibility in how they're executed. Most marketing tasks fall into this category. For instance, drafting social media posts, writing emails, or creating outlines for blog articles could be considered “good-enough” tasks. Define clear criteria for what “good enough” looks like for each task.

Expert Tasks: These are at the top of the hierarchy and rely heavily on context, experience, and creativity. They're the hardest to delegate to AI because they often require nuanced understanding and decision-making. While AI can sometimes handle expert tasks, it often requires extensive context and may be less controlled.

First, focus on delegating objective and good-enough tasks to AI, as these are easier to automate effectively. Then, try breaking down expert tasks into smaller, more manageable subtasks (objective and good-enough tasks).

#3: Create AI Playbooks for Task Execution

A well-structured playbook allows you to train AI to perform specific tasks consistently and effectively, much like you would train an intern or new team member.

Your AI playbooks should include several key elements: 

Context and mindset: This section provides the high-level framework and goals for the task. This helps the AI understand the broader purpose of what it's doing. For example, in a social media response playbook, you might explain the brand's voice and overall customer service philosophy.

Specific Steps and Instructions: Here, you'll outline the process in detail, breaking down the task into smaller, manageable steps. Be as clear and specific as possible, as if you're explaining the process to someone who's never done it before.

Examples of Good Work: Provide multiple instances of what high-quality output looks like and examples of what to avoid. This helps the AI understand the nuances of your expectations. In the case of social media responses, you might include a variety of well-crafted replies to different types of comments. 

Specific Criteria for Success: Clearly define what constitutes a successful outcome—measurable standards the AI can use to evaluate its performance. This could include specific metrics, qualitative aspects, or a combination of both. For social media responses, criteria might include response time, tone consistency, problem resolution rate, and customer satisfaction.

Specific Prompts: These are the exact instructions you'll give the AI to execute each step of the process. Refining these prompts over time can improve the quality and consistency of the AI's output.

While playbooks can be lengthy (often 10+ pages), start small with your most important processes. You can expand and refine them as you see what works best. 

Writing Effective AI Prompts for AI Operations

To get the best results from AI tools, you need to write clear and effective prompts. Use the “MASTER” method to create prompts that yield high-quality outputs.

Markdown: Structure your prompts using Markdown formatting. This helps organize the information in a way that's easy for both humans and AI to parse. For example, use headers to separate different sections of your prompt and bullet points to list specific instructions.

Act as: Tell the AI to act as a specific role or persona. This helps frame the task and can improve the relevance of the AI's responses. For instance, you might instruct the AI to “Act as an experienced social media manager” when asked to draft posts.

Specific: The more detailed you are about what you want, the better the AI can deliver. Instead of asking for “a blog post about marketing,” specify the topic, desired word count, target audience, and key points to cover.

Thread: Consider what's in the conversation thread. In tools like ChatGPT, the AI considers the entire conversation when generating responses. Be aware of this context and how it might influence the AI's output.

Examples: Showing the AI samples of what you're looking for can dramatically improve the quality of its output. If you're asking for a certain writing style or format, provide an example of what you consider ideal.

Regenerate: If you're unsatisfied with the AI's output, don't hesitate to ask it to try again.

Pro Tip: Clicking on the Regenerate symbol next to the AI's output adds randomness to its response. Instead of prompting it again, refine your original prompt based on what worked or didn't in the initial response. Then, you have a proven prompt to store in your playbook.

#4: Give Feedback to AI Systems

Improving your AI operations is an ongoing process that requires active feedback to refine the outputs. This feedback loop is crucial for getting the best results from AI tools over time.

One effective method is to ask the AI to critique itself. You can instruct the AI to evaluate its output based on the criteria you've provided, which can help identify areas for improvement. This self-reflection process can often yield insights into how to refine your prompts and instructions.

Creating a scorecard to evaluate the AI's performance can also be helpful. Define specific metrics or qualities you're looking for in the output, and rate the AI's performance on each. This provides a structured way to track improvements over time and identify consistent areas of strength or weakness.

Providing examples of what “good” looks like is another powerful feedback method. If the AI's output isn't quite meeting your standards, show it examples of what you consider high-quality work. This gives the AI a concrete target to aim for in future attempts.

Again, think about this process as if you're training a new employee. The more guidance and feedback you provide, the better the results will be over time. It's an iterative process that requires patience and consistent effort, but the payoff is a steadily improving AI assistant that can handle more complex tasks with greater accuracy.

Other Notes From This Episode

• Connect with Michael Stelzner @Stelzner on Facebook and @Mike_Stelzner on X.
• Watch this interview and other exclusive content from Social Media Examiner on YouTube.

Where to subscribe: Apple Podcasts | Spotify | YouTube Music | YouTube | Amazon Music | RSS

✋🏽 If you enjoyed this episode of the AI Explored podcast, please head over to Apple Podcasts, leave a rating, write a review, and subscribe.