A co-pilot approach to genAI (with prompt examples)

The relationship between marketers for-2024-and-beyond">and generative AI has been complicated from the start. Many marketers have approached AI-generated content with understandable skepticism, questioning its ability to capture the nuanced voice and creativity that great marketing demands. Below the surface of these concerns is the fear that AI might eventually replace human marketers altogether.

Yet these concerns, while valid, need not define our relationship with AI. A better way is to approach it as a co-pilot. Marketers can harness generative AI to enhance their capabilities while maintaining creative control and strategic oversight.

Let’s explore how to implement this co-pilot approach across three essential marketing tasks: 

• Crafting compelling blog posts.
• Creating engaging imagery.
• Developing effective social media content.

Let’s examine how AI can amplify your creativity and productivity and aid the human insight and strategic thinking of exceptional marketing.

Training generative AI

Think of generative AI as a brilliant new hire who graduated at the top of their class but has never worked in your industry. They have good general knowledge. However, they lack important context about your brand, customers, and industry that comes from experience. You must educate your AI co-pilot about your company’s brand voice, target customers and industry.

When you train your AI assistant, don’t share sensitive business information or trade secrets. Instead, focus on public-facing content that showcases your brand voice and industry expertise. This transforms it into a specialized marketing partner that understands your company’s unique perspective and communication style.

Creating your custom AI co-pilot

OpenAI and Anthropic offer different approaches to creating trained AI assistants. OpenAI has Custom GPTs Anthropic’s Claude has projects. 

Below is the process of creating an assistant using OpenAI’s Custom GPT capability.

• Navigate to the GPT Editor at chatgpt.com/gpts/editor.
• Craft your training prompt and include:
  • The GPT’s primary purpose and goals.
  • Essential knowledge areas for your industry.
  • Target audience description.
  • Instructions for analyzing training materials.
  • Guidelines for maintaining brand voice.
• Upload approximately ten content pieces that demonstrate your brand voice and style. These might include:
  • Blog posts.
  • Marketing emails.
  • Social media content.
  • Website copy.
  • Brand guidelines.
• Give your custom GPT a memorable, relevant name that reflects its role in your marketing process.

Here is an example training prompt:

“This GPT helps marketers at AI training and consultancy firms. It assists them in creating high-quality content. The content is aimed at marketing directors. These directors want to use AI to improve business results.

The GPT knows a lot about AI. It understands generative AI tools like ChatGPT and Claude. It also knows about machine learning, marketing analytics, and Agile Marketing. This includes the values of the Agile Marketing Manifesto. It understands the challenges and goals of marketing directors at mid-sized to global Fortune 2000 companies. This includes expanding initial AI efforts, finding high-impact use cases, training teams, and managing organizational change.

Responses are credible, accurate and conversational, tailored to marketing professionals who are familiar with AI but seek practical guidance. They write in a welcoming, professional brand voice based on the attached examples.”

I also give the GPT a list of phrases and words to avoid. My source for those phrases and words are:

• GPTZero’s AI Vocabulary.
• “Top Gobbledygook phrases and how to avoid them” by David Meerman Scott. 

Using AI for research

Establishing a solid foundation of knowledge is crucial before creating content — even for topics you’re familiar with. AI-powered research tools, particularly Perplexity.ai, help us gather and validate information by providing source-backed insights and generating intelligent follow-up queries.

Start your research journey with broad, open-ended questions that help map the territory. For instance, asking, “What do you know about the use of AI to improve marketing analytics?” opens up multiple avenues for exploration. AI can quickly synthesize information from various sources while suggesting angles you might not have considered.

Reviewing the sources and exploring suggested queries, your understanding deepens and naturally evolves toward more specific questions. Take notes during this process, paying particular attention to:

• Emerging patterns and themes.
• Surprising insights or counterintuitive findings.
• Recent developments in your field.
• Potential gaps in current industry discussions.

This iterative approach helps you develop a unique perspective on your topic. It lets you move beyond surface-level observations to insights that will truly engage your audience. Those with deep expertise in a subject might be tempted to skip this research phase. However, even a brief AI-assisted exploration can reveal angles and new developments that enrich your content.

Here are some sample queries that might help:

• What information would help a marketer understand how to enhance their marketing using AI and marketing analytics? What information would help my target marketer do her job?
• What research studies about AI and marketing analytics would support the average B2B marketer? Provide references for each research study.
• What do marketers hate most about marketing analytics? How can AI help solve some of these problems?
• What widely accepted best practices about marketing analytics are not actually that helpful?
• What seemingly unrelated topic or field has crucial lessons for the use of AI in marketing analytics?

Dig deeper: Meet my research team: Gemini, ChatGPT and Perplexity

Creating an outline for your blog post

Your blog post begins to take shape in the outline phase where generative AI can be invaluable. However, the key to success lies in how you communicate with your AI co-pilot. Let’s explore a structured approach combining a proven framework and strategic thinking.

The RAPPEL framework

Trust Insights’ RAPPEL framework provides an excellent structure for crafting effective AI prompts. Let’s break down each component and see how it applies to blog outline creation:

Role: Define your AI’s expertise and voice. Be specific about the knowledge domains you want it to draw from. You will have done some of this already in your Custom GPT (see above). But you may want to add additional context. For example:

• “You are a data scientist with expertise in marketing analytics and AI. You are skilled at explaining complex topics at an 11th-grade reading level and excel at creating engaging, shareable long-form blog posts.”

Action: Clearly state what you want the AI to do:

• “Create an outline for a long-form blog post, focusing on structure and key points before we move to content creation.”

Prime: Gauge the AI’s knowledge and establish context:

• “What do you know about applying AI to marketing analytics and enhancing campaign ROI? What are current SEO best practices for Google and Bing?”

Prompt: Deliver your specific request with detailed parameters:

• “Create an outline for a long-form blog post about using AI to enhance marketing analytics. Target audience: marketing directors in mid-sized to Global 2000 firms. Include sections on fundamentals, data analysis, pattern recognition and real-world case studies. Focus on SEO keywords ‘AI marketing analytics’ and ‘AI-based marketing.’”

Don’t be satisfied with the first outline you receive. Iterate on that outline to refine it. Try using other models (Claude, if you start with ChatGPT, for example) and asking it to review and refine your outline. Check that the outline meets your SEO goals, aligns with your audience and provides sufficient depth and uniqueness.

ChatGPT’s 4o model with Canvas can be particularly helpful for refining your outline. It allows you to make changes to the output without copying and pasting.

Best practices for outline development

• Iterative refinement
  • Review the AI-generated outline critically.
  • Add missing sections that reflect your expertise.
  • Remove redundant or off-target elements.
  • Ensure logical flow and progression of ideas.
• Multi-model validation
  • Use different AI models to review and enhance your outline.
  • Compare perspectives from various AI tools (e.g., ChatGPT vs. Claude).
  • Synthesize the best suggestions from each model.
• SEO integration
  • Verify keyword placement in key sections.
  • Ensure headers align with search intent.
  • Plan for internal and external linking opportunities.
• Audience alignment check
  • Confirm each section addresses audience pain points.
  • Verify the technical depth matches the audience’s expertise.
  • Plan for progressive disclosure of complex concepts.
• Content depth planning
  • Mark sections requiring original research.
  • Identify opportunities for data visualization.
  • Note areas needing expert quotes or case studies.

Learn: The final “L” in RAPPEL stands for “Learn.” This is your opportunity to create a feedback loop that improves future content creation:

• Ask the model to build a prompt that encapsulates the task.
• Document successful prompts in a prompt library.
• Tag prompts by content type and goal.
• Note which AI suggestions worked best for your audience.
• Record patterns in outline structures that drive engagement.
• Save templates of high-performing outline formats.

Remember, AI can help you create a good outline. However, your knowledge and understanding of your audience should shape the final structure. Let your strategic insight determine what is needed to resonate with readers and achieve marketing objectives.

Writing and revising the blog post

The journey from outline to polished post offers two viable paths. You can either craft the first draft yourself, or collaborate with your AI co-pilot to generate an initial draft. Both approaches have merit — choose the one that best fits your writing style and time constraints.

Using AI for revision

AI can help you revise your blog regardless of how you create your first draft. Consider these strategies:

• Use multiple AI models to gain diverse perspectives on your content
• Ask for specific feedback on elements like:
  • Clarity and flow.
  • Engagement factors.
  • Technical accuracy.
  • SEO.
  • Call-to-action effectiveness.

Remember, AI serves as your assistant, not your replacement. The final decisions about voice, tone and content should be made by you and align with your expertise and brand standards. Let AI enhance your writing while maintaining the authentic human touch that connects with your readers.

Dig deeper: 9 best practices for AI tools in content creation

Creating imagery for your blog post

Now that you have your blog post, you need to generate one or more images for it. I use Midjourney because it generates the most photorealistic images. Here is a sample Midjourney prompt:

“Create a natural, candid photograph of a team of marketers standing in a modern office space with soft, natural daylight streaming through floor-to-ceiling windows. They are studying marketing analytics on a large digital dashboard displaying data and graphs. The room has warm wood floors and potted plants that add organic elements. Genuine expressions of engagement and casual professional attire are captured. 

Taken with a 35mm lens, shallow depth of field, forming a cinematic composition. Style notes: a photojournalistic approach, natural color palette, soft directional lighting, environmental context and authentic human interactions with widescreen cinematic framing. 

–quality 2 –sref http://marketingfrontier.ai/wp-content/uploads/2024/10/Identifying-Use-Cases.png –aspect 16:9″ 

Here, I provided very specific instructions, including style hints. I’ve also asked it to:

• Take more time (the “-quality 2” command).
• Use a style reference image (the “sref” command with a publicly available link).
• Use a particular aspect ratio (the aspect command).

Most of this you can do in the Midjourney interface. It will then give you a choice of four images. You can choose one you like and ask it to create variants of the image using either the “vary subtle” or “vary strong” commands.

Above is the image that Midjourney created. It’s not perfect, but it’s good enough for my needs.

Another tool I use is Napkin AI. It transforms text into compelling, McKinsey-like visuals. 

Dig deeper: The new frontier of visual content: A marketer’s guide to AI

Creating social media posts

Now that you have a finished blog post and an image for the blog post, it’s time to promote it on social media. Here’s a prompt for creating social media posts:

“Role: You are an expert social media content creator specializing in crafting engaging, educational and semi-casual posts for LinkedIn, X and Facebook.

Task: Promote the blog article found at https://marketingfrontier.ai/optimizing-ai-content-marketing-with-machine-learning by creating LinkedIn, X and Facebook posts that increase engagement and educate the audience. Each post must be optimized for its platform, catering to LinkedIn’s professional audiences, the typical marketer who uses X and Facebook users who are marketers. 

Specifics:
Tone: Semi-casual
Content: High-level insights about AI and marketing
Keywords: AI Content Marketing, Machine Learning

The goal is to make the posts engaging and enjoyable while being informative.

Here are some examples of the kind of social media posts I like.

Example 1:

“Outside of maybe Steve Jobs, Jensen Huang of Nvidia is the most impressive CEO I’ve ever heard speak. His presence and approachability. His humility. His command of details in every aspect of his products and industry. His confidence in his business. His authenticity. His respect for his peers and competitors. And his unwavering belief is his vision for the future.

He has an unparalleled view into the current state of AI and he is literally building the future of AI (and the world) with his GPU-powered data centers and accelerated computing platform. 

He recently sat down for an extensive interview with Brad Gerstner, founder and CEO Altimeter Capital and co-host of the BG2 podcast. 

I highly recommend reading the entire post, but here are some of the most interesting excerpts:

> On why he believes AGI is coming soon: “A lot of this is happening because we drove the marginal cost of computing down by 100,000x over the course of 10 years.”

> “The data center is now the unit of computing.” This has profound implications to the future of their business and how AI labs are thinking about building next-gen models.

> “We’re not trying to take any share from anybody. Nvidia is a market maker, not share taker. If you look at our company slides, we don’t show, not one day does this company talk about market share, not inside. All we’re talking about is how do we create the next thing? What’s the next problem we can solve?” I love this perspective in business.”

Mike and I talk a lot more about the interview on Episode 119 of The Artificial Intelligence Show, which drops Oct. 15.”

Example 2:

“Here is why I think we are underestimating the impact of AI over the next decade: our projections are based on running faster, cheaper and with less people, rather than fundamentally reinventing what we do and how we do it. #AI #innovation #leadership #disruption #future”

Ensure each post is no more than 200 words.

Use a conversational and relatable tone to connect with the audience.

Include relevant hashtags and a call to action to boost engagement.

This task is vital to my career. By creating these posts, I aim to increase engagement and educate my audience, which is crucial for building brand awareness and attracting new clients.”

This is a long prompt, structured to begin with the role, followed by the task (action), specifics about tone, content and keywords, examples and a clear emphasis on the task’s importance to my career.

Once I have the results from this prompt, I typically revise it and schedule it on the appropriate social media platform.

Embracing AI as your marketing co-pilot

AI works best not as a replacement but as a skilled assistant that strengthens your marketing. With clear communication and regular training, you can create an AI partnership. This partnership will improve every step of content creation, from research to social promotion.

You might think using my process would take longer than your current process. In my experience, this isn’t the case. It takes less time because you control the entire process rather than depending on other groups for imagery or social media. It is also true that many of us short-change the process we know we should do in the interest of time.  

For example, we don’t necessarily check for audience fit. We are in churn-and-burn mode.  With AI, we can put out higher-quality work in the same or less time because AI is doing some of the basics.

Consider this: 

• How might your marketing role evolve when your AI co-pilot handles routine tasks? 
• What creative projects could you tackle with your newly freed time? 
• What unique perspectives could you bring to your content when AI handles the heavy lifting of research and initial drafts?

The future of marketing is not about humans versus machines. It is about humans and machines working together. This teamwork will create more engaging and impactful content. The question isn’t whether to adopt AI but how to make it one of your most valuable team members.

Dig deeper: 3 marketing use cases for generative AI that aren’t copywriting

The post A co-pilot approach to genAI (with prompt examples) appeared first on MarTech.