How can PMs benefit from generative AI

Over the last two years, developments in generative AI have accelerated the pace of work and all indications point towards this continuing into the future. PMs now can:

• Create an interactive prototype in less than a minute
• Synthesize customer interviews in a few seconds
• Define solid product experiments for assumptions
• Receive feedback from multiple perspectives without talking to anyone
• Produce authentic images with simple prompts

These improvements have made people begin to question whether AI will eventually make PMs obsolete. However, I remind them that while AI can amplify a product manager’s potential, it still needs them to act as the pilot.

Keep reading to learn about how you can use AI within your role as PM by understanding its tools and how to prompt it.

Using AI as a co-pilot

In “Reimagined: Build Products with Generative AI,” Shyvee Shi emphasizes the importance of using AI as the co-pilot, not the pilot. But what does that mean in practice?

You’re the driver, not the passenger.

AI helps you solve the challenges you face faster than you would alone, but that implies a few critical things:

• Expertise — You’re the expert in your area, so you should use your knowledge to advance and accelerate
• Questions — Knowing which questions to ask and how to ask them will considerably impact the quality of the answers
• Decision-making — As the driver, you need to call the shots, know where to go, and choose where not to go
• Problem framing — It’s fundamental how you provide context when using AI. The more precise you frame the problem, the more trustworthy the outcome will be

Being the driver means you know your field. AI can help you when used correctly. However, it may distract you when misused.

Understanding the power of prompts

When you write well, you can express your thoughts concisely and consistently. This has become even more important because generative AI uses prompting at its core.

Knowing how to use prompts shapes the outcome you receive. A good prompt consists of:

• Desired expertise — Define how you want AI to behave (e.g., product coach, agile coach, UX designer)
• Context — Elaborate on the context as straightforward as you can
• Task(s) — Define the tasks you want the AI to perform for you
• Output format — Clarify how you want the results to be presented (e.g., table, bullet points, etc)
• Level of detail — How detailed should that be? You can use a few tricks, such as “Explain it to a five-year-old.” “Explain it to a software engineer.” “Explain it to a 50 year old”

Now, let’s look at a few examples of good and bad and see what ChatGPT can do for us.

Over 200k developers and product managers use LogRocket to create better digital experiences

Learn more →

First, imagine you put in:

“Analyze the attached file and give me tips on what to write about.”

The prompt fails to provide enough details to get a valuable answer. In this case, GPT leaves you with:

Now, imagine you strengthen the to prompt:

“Act as an experienced author in the field of digital product management. Given the answers from the attached file, I want you to: 1. Identify three strong patterns that readers long to understand 2. For each pattern, define a perspective to write from, points to pay attention to, opportunities that would attract readers 3. Present the results in a table”

Now, the prompt returns a far more useful response:

Generative AI tools

Generative AI tools enable you to autonomously create outputs like text, images, videos, and other artifacts. Given the proper context, the application can produce the desired results. Here are three examples of how you can use generative AI tools:

1. Creating designs without product designers

Uizard.io, recently acquired by Miro, empowers people to supercharge how they create designs. I used a prompt to create a platform for product managers to build community:

Crafting an interactive prototype took roughly one minute. I wouldn’t call the result final, but it’s a place to start and make it better.

I use Uizard.IO to accelerate early-stage idea testing:

2. Writing product briefs from customer interviews

During an exercise with one of my cohorts, I asked participants to practice interviews and collect what they earned. They needed to consolidate the key points. It generally takes 10-15 minutes, but AI can do it in less than a minute.

Miro recently added built-in AI functionality to the intelligent canvas; one feature enables users to synthesize research or write product briefs quickly:

I used the “product brief” functionality and got the following result in about 45 seconds. I found it quite insightful and a great starting point. Writing at this level would take an experienced PM at least 20 minutes.

3. Writing tech backlog items from page insights

Have you ever heard about Page Insights? If not, it can help you understand the critical points of your website, desktop, and mobile version and improvement opportunities. Yet, that may be pretty technical for PMs.

I tried analyzing Substack and exploring potential opportunities:

I use the following prompt on ChatGPT to analyze the exported PDFs and create backlog items:

“Behave like a senior product manager and analyze the attached files related to Substack page insights. Identify potential opportunities and create product backlog items. For each item, describe what it’s about, why it matters, how complex it is, the benefit of doing it, the priority, and how to explain that to a software engineer. Present the result in a tabular format.“

This produces tangible insights that you can use to drive product decisions:

What can’t AI replace?

Although AI has a lot of potential, it still fails to compete against you on the following points:

• Human judgment — AI equips you to make decisions, but it’s your job to call the shots, not AI
• Data quality — It’s crucial to provide trustworthy data to whatever AI tool you use. With poor data, you’ll receive incorrect results presented in a compelling AI. AI shows outputs with high confidence, which could trick you if you have faulty data
• Empathy — Nothing can replace real contact with customers. Many of my learnings as a product person happened when I observed people performing a job we wanted to help them with. Empathy enables us to connect as humans, and I don’t see AI replacing that

Use AI when it’s the best solution for a problem, but refrain from using it just because of its hype.

Ask questions like, “How might we solve this problem?” Don’t ask questions like “How might we solve it with AI?” The first is open for exploration, while the second may lead you to implement a misfit solution.

Key takeaways

AI amplifies your potential when you use it as a co-pilot. However, don’t forget you’re the driver, not the passenger. Even with generative AI tools, you still need to provide relevant context and problem framing.

Also, it’s important to remember the limitations you face with AI. As a human you have empathy that a computer cannot easily replace. Alongside this, you still need contact with your customers to understand the problems they face. The best strategy combines your strengths as a PM with the productivity boosting powers of AI.

Featured image source: IconScout

LogRocket generates product insights that lead to meaningful action

LogRocket identifies friction points in the user experience so you can make informed decisions about product and design changes that must happen to hit your goals.

With LogRocket, you can understand the scope of the issues affecting your product and prioritize the changes that need to be made. LogRocket simplifies workflows by allowing Engineering, Product, UX, and Design teams to work from the same data as you, eliminating any confusion about what needs to be done.

Get your teams on the same page — try LogRocket today.