By: Alan A. Spina

The disruption generated by the launch of ChatGPT added to the consequent exponential advance of artificial intelligence towards the end of 2022 and resulted in a democratization of the use of artificial intelligence, generating a series of opportunities as well as risks and serious dangers even still unknown, mainly regarding good practices in the use of these tools.

The benefits proposed by AI consequently tempt those users in different professional fields, with the role of the Product Manager (“PM”) not being alien to this situation.

It is fascinating that through the more than simple use of prompts, it is possible to facilitate or even replace some of the somewhat operational tasks carried out by PMs, who, when it comes to wanting to replace and/or use these tools, encounter similar difficulties as the devs when developing using AI as the sole and exclusive source. The answer to this problem is that this tool is not sufficient and does not fully resolve the user’s needs.

This does not mean that these tools do not serve, help, or even improve the creative process that PMs develop, but as their name indicates, they are tools. Therefore, they should not be used as the sole and exclusive source of consultation and should not be considered able to replace the role automatically. Care must be taken regarding its use.

Just as in my case, I do not usually recommend using AI for junior developers (since they can cause spaghetti code or not be in accordance with the software architecture of the solution you are working on). I also do not recommend the indiscriminate use of AI for PMs, and this does not mean that artificial intelligence is not a good resource and consultation or guide tool that supports the PM to improve and make their role more efficient. This is mainly regarding the creation of documentation as well as history stories.

My primary purpose is that today, we can focus on the analysis of how AI can help us make writing more efficient and improve user stories, where the operability of the PM task is manifested with greater emphasis without falling into the easiness of believing that AI has all the answers to the needs of our clients.

The criticality of user stories in the development process is more than evident since they are not only essential for any PM, being a large important part of their role, but they also help the entire work team, but above all, developers, to correctly identify the interpretation of the scope of business needs.

Among the benefits that AI proposes, I consider that it undoubtedly helps PMs improve and optimize user stories while improving reliability, scope, and scenarios. However, we should not confuse these benefits with waiting for ChatGPT or any of the artificial intelligence available for consultation work magic to solve the writing of all the documentation required in specific projects with very particular and specific needs. The differentiation in this is marked by a responsible PM who will quickly realize that the information is not correct or sufficient to respond to the needs of the business or the end user.

Let’s go with an example

Let’s see an example of creating a user story for the sign-up of a casino page.

Link on ChatGPT:

Let’s be simplistic and ask Chat GPT (3.5) to complete a user story in one line:

The result is amazing! ChatGPT gave us everything we needed where we could hand this user story to the developers to do the work and go off to do other tasks. But not so fast, let’s analyze it in detail:

Generic user story. Lack of scenarios

Are we sure this is the only place we want to see the Sign Up button? Is it the only flow or scenario that would take the user to Sign Up? Having worked in the field, I know that the most important thing is for the user to bet on the page, which is why many pages allow users to bet without having a user and create a user associated with the “guest” account with which they bet. In this case, ChatGPT gives us only one scenario of the many that can occur (and generally the most generic) to sign up on a casino page.

Lack of information

The information that ChatGPT provides is incredible. It even helps to remember that you need to validate the age of the person who wants to play, but that does not mean we need, for example, the day of birth. A solution would be to just ask them if they are of legal age or not to avoid having to use a date picker and thus avoid more steps. Here, we would undoubtedly be losing useful information that could provide us with a more detailed user profile when, for example, routing MKT actions aimed at said specific user, but it is up to the PM to make the decision of what is most important (being able to resolve with an A/B test, if warranted).

It is also important to remember some details regarding the request for information due to the legal restrictions that we must consider with the rise of personal data regulation, mainly in the EU as well as in some countries such as Brazil and Italy that have advanced regulations in terms of protection of users’ personal information. Not to mention, if we consider that the field of online betting could be prohibited in some countries and could fall into illegalities in the case of not considering requesting information such as the address of the users or the country from where the bet or game is being made.

Excess steps

These steps are correct for most web pages, but in the specific case of casino games, what is expected is that the user bets as quickly as possible, and to do so, they do not require the email to be verified or the data to be validated. For example, on many casino sites, you can enter an email incorrectly and in case of an error, they have 24/7 support to resolve user problems.

Conclusion: Use GPT and AI to our advantage

AI tools are undoubtedly here to stay, and we cannot deny, as I have already mentioned, that in some way they come to help PMs do their work, as Marty Cagan, founder of Silicon Valley Product Group, says:

“… But when it comes to the impact on product managers, I have real hope that the combination of product manager’s informed product sense and judgment, with new and very powerful tools to challenge and augment our thinking, will result in making better product decisions faster.”

The tools are here to help PMs do their jobs faster and better; the question is how to do this.

Collaborate with AI tools

If you don’t use AI tools and want to start using them, do your work as usual, show it to an AI and ask it to rewrite it or give you advice on what to improve. Use the tools as an aid to your work. Ask scenarios, options, alternatives, people, ways to monetize, survey questions, and journey maps, and you will be surprised by the capabilities of AI for your day-to-day work. But remember that the last word is yours and that AIs do not know the entire context of what you are working on, nor do they know the user and the problem you are working with. For the latter, there are PMs.

Remember the contexts

Don’t forget about the contexts. It’s very important; it’s not the same as writing, “Write the user story for a feature that creates a playlist based on your horoscope,” or “Act as the Product Manager of Spotify. Imagine you were just creating a playlist feature for Spotify, which collects songs based on your horoscope sign. Write the user story for that build.” Look at the differences in the outcomes; they will be surprising and, above all, more precise.

Rewrite when necessary

Generative AI is excellent in its way of writing and improving the way of saying things, but just because it is written in the best grammatical way does not mean that it is the easiest or correct way to write a story of a user. Remember that user stories are so developers and the entire team can easily understand what needs to be done and the scope of the tasks. They are not literary documents or documentation for end users or clients.

Don’t go overboard

User stories produced with artificial intelligence generally provide many alternatives and actions or features that are outside the scope of the task. Although all product managers want to have all the functionalities in their product, we must understand the importance of the scope of what we want to develop and avoid falling into the trap of going beyond what is necessary.

Let’s not forget the technical aspects

In general, user stories will not have a technical focus, but this does not mean that it is not important to include it in them. Let’s ask artificial intelligence the same questions that we will ask our technical leaders to have a prior approach when discussing the user story with them.

Test tools

We are in the AI ​​boom. There are new tools every day that tempt their users to their daily and indiscriminate use. I consider it essential that you try several of them and see which ones work best for you. Compare results with Gemini, ChatGPT, Copilot, or Meta. Other tools exist, such as specific ChatGPTs for creating user stories, or you can create your own ChatGPTs with your company information to increase their context. There are tools like, which are AI specialized in user stories, but again, they are not going to do the work for you rather, they are going to help you do it better or faster

Avoid incorporating personal and confidential client information

Remember that these tools are collaborative. They use the information the user provides to learn, which is why we could incorporate confidential information from our clients, potentially putting new projects or secret developments at risk and legally exposing the company.


AI is a companion that is here to stay and help improve and optimize the role of the PM, but it is not everything. Let us remember that the main characteristics that describe and highlight a good PM are his curiosity and his complete and deep knowledge of the business and the needs and profile of the client. Considering that an AI can replace curiosity, creativity, and human sensitivity is a very serious conceptual error that can generate delays, losses, and financial impact on the client. 

My recommendation is that we use it consciously, knowing that nothing replaces human contact. What differentiates us from machines, at least until today, is that we have the ability to feel and think for ourselves.