How to Write User Stories Using ChatGPT

Hi, this is Helena, and welcome back to my YouTube channel. I have such a great tip for you today. I am going to share with you how you can write your user stories, epics, acceptance criteria automatically using ChatGPT and AI. I know it's a little bit mind-blowing, but it just— you have to stick with me for the entire presentation, and I'm going to show you exactly how to do this. Now, I know AI is Such a hot topic right now, but not many people know how to actually apply it into situations that can help them become more productive. So I have figured out the prompts that you can use in AI to help you write user stories and help you become more productive as an agile project manager. So for those of you who are new to agile, let me explain how it works. So if you're managing a project using Scrum or Agile, then what you will have is a product backlog, and the product backlog would include all the requirements. For your project. So essentially, when you complete all of the requirements in the product backlog, you would finish the entire project. Now, all of these requirements in your product backlog would be expressed in the format of a user story because you want to become customer-centric. So for your projects, you always want to keep the customer at the front and center of your minds. And once you have all of your requirements written out, you will meet with the entire team. So you will meet with the Scrum Master, the Product Owner, and everyone on the development team. Everyone would get together to talk about all of the user stories that you have put in the product backlog, and you would talk about which ones are more important, which ones are less important, which ones are going to take more time, less time, which ones you need more clarification on, et cetera. So at the end of that meeting, you will have a prioritized backlog with the most important user stories at the top and the least The important ones at the bottom. And then you're going to go ahead and say, okay, during our next sprint, how many of the ones at the top can we complete? And you say, okay, we can probably complete the top 3. Then you would take the top 3 and put that into your sprint backlog, and that's what you're going to be working on during the next sprint. And usually a sprint is 2 weeks long, but depending on the company, it could be anywhere between 1 to 4 weeks long. Generally, the rule of thumb is that it's 2 weeks long. So now that being said, we're going to talk about how can you like use AI to help you write these user stories much faster and maybe even discover some of your blind spots that you didn't think about before. So if we log into ChatGPT here, and the thing is not many people know that you can actually use ChatGPT to generate different voices. So if I say write a financial report in the voice of Jim Cramer, that would sound a lot different if I asked ChatGPT to write a financial report in the voice of Barack Obama or something, right? So in this case, I want ChatGPT to act as a world-class product owner. Generate epic— so epic is a group of related user stories— user stories, acceptance criteria, and tasks for the following software feature. All right, so let's suppose, you know, I'm just going to use a random example right now, that we're working for Amazon and the feature we're trying to develop is the refund feature for their customers. So it's refund for customers who are not satisfied with the product they purchased on our e-commerce website. Great. And you can even put, you know, Amazon or whatever the name of the site is. And we're just going to give it a minute for ChatGPT to finish generating here and see how fast it is writing, right? I don't believe any human can write as fast as what ChatGPT is doing here. And the response is actually not bad. What you could also do is just copy everything and then have ChatGPT regenerate the responses. So if you do go ahead and press this button right here, all of the content that I just generated will be erased. So if you like, let's say, this user story but not the next one, then what I would recommend is like Just open a Google Doc, copy and paste the response here in there, and then press regenerate. Let's take a look at what ChatGPT here have written. As a product owner, I understand that implementing a refund feature for customers who are not satisfied with the product they purchased on our ecommerce store is very important to customer service. So you probably don't need all of that when you're actually writing. This epic focuses on developing a robust and efficient refund management system that ensures a seamless experience for customers who wish to return a product and receive a refund. So that's pretty good. This is a good summary of the epic. Now let's take a look at the user stories that ChatGPT generated, and it generated 3 different user stories. So the first one is, as a customer, I want to be able to initiate a refund request for a product I'm not satisfied with so that I can receive a refund for my purchase. Okay, and the acceptance criteria. So acceptance criteria are things that the developer needs to know more about in order to see how do they want you to actually implement this specific user story. Let's say it was a login feature that we're working with. And, you know, you can just say like, as a customer, I want to be able to log in to the store so that I can complete my purchase or something like that. So if a developer just sees that, they could imagine 10 billion different ways to get that done. Is the username and the password button up on the top right-hand Or the bottom left-hand corner? And do you need a forget password somewhere in there? What happens once they press that button? Is it going to their dashboard? Is it going to the homepage of the website? You know, there's so many things that could happen that it's very vague if you just give a developer just a user story, because there's so many ways that they could complete that user story, and so many things could be left to interpretation. And that's why— Every user story also has a corresponding set of acceptance criteria that tells you exactly how you want the developer to implement this user story. So the user story, you can think of it as why do we need this feature, and then the acceptance criteria you can think about as how are we actually going to complete this feature. So every user story would need to have its own corresponding acceptance criteria that goes along with that. So in this scenario here, we can see the acceptance criteria that AI has written for us are: the refund request should be initiated through the customer's account on the e-commerce site. The customer should be able to provide a reason for the refund request. The system should validate if the refund request is within the allowed timeframe for returns based on the return policy. Important, right? The system should generate a refund confirmation for the customer and notify them via email. So those are all very valid acceptance criteria. The one thing I should mention here is that you can see like AI is not just pulling things from the internet and regurgitating it to you. The way that AI actually works is that it's mimicking the way that our brains work, and our brains work by association, by neural linking and networks. And the people who made this technology tried to mimic the way that our brain worked into this technology. I know it sounds creepy on some levels, I get it. So most people think like maybe AI is just a big database stored somewhere and it's just pulling information as you're typing here, but that's actually not the case. There is no database. There is no database. And this fact really blew my mind as I got to learn more and more about AI. So what's actually happening is learning about these different topics and it's actually assimilating, like literally learning and then being able to give it back to you, regurgitating it back to you in a way that made sense, Based on what your command is and is doing this in a matter of seconds, which is just completely mind-blowing. All right, now let's take a look at the second user story that AI wrote here. As an admin, I want to be able to review and approve or reject refund requests so that I can ensure the validity of the refund request. Totally valid user story here. The refund request should be listed in an admin dashboard with details of the customer product, reason for refund, and refund amount. Yes, we definitely do need that. The admin should be able to review refund requests and verify if it meets the return policy criteria. So as you can see, like it's actually doing a pretty good job here. And then in the end it has some tasks here. So when you keep continuing to write in ChatGPT, what it's actually doing is going to read everything above and generate more responses for you. So I can say, Please generate more user stories for this feature. Right. And then we just continue, right? As a customer, I want to be able to track the status of my refund so that I have visibility into the progress of my refund. Another one here, as an admin, I want to be able to generate reports on refund requests and their status so that I have insights into refund trends and performance metric. As a customer service representative, I want to be able to communicate with customers regarding their refund requests so that I can provide timely and accurate updates. As a customer, I want to be able to provide feedback on the refund process so that I can share my experiences and help improve the refund process in the future. So you can see it's actually generating— it can just keep going and going. If I just wrote, please generate more user stories for this feature, but it was a brand new chat, then ChatGPT would have no idea what kind of user story I'm asking it to generate for. But whenever it's writing, it's actually reading all the text above and then looking at everything out on the internet, learning, and then giving you the information. And it's doing this in literally in a matter of seconds. Like, it's just pretty incredible. What this technology can do and how much it could help you speed up your productivity, right? So here's some more responses it just generated here, some more user stories here. As a customer, I want to be able to cancel my refund request if I change my mind. As an admin, I want to be able to analyze the reasons for a refund request so that I can identify potential product or service issues and address them proactively. So you can see like it's actually learning and generating pretty great responses. If you find that, you know, ChatGPT isn't generating the type of response that you want, you could get more specific. So at the top here, I just said act as a world-class product owner, generate epics, user stories, acceptance criteria, and tasks for the following feature. And I just wrote one line for this feature. So if you want ChatGPT to get more specific in your prompt, you can add in more details about the feature that you want. It to generate. All right, so this pretty much wraps up today's lesson. I hope you learned a lot. Keep these 2 prompts in mind. So the first prompt here, just to recap, is, as a world-class product owner, generate epics, user stories, acceptance criteria, and tasks for the following feature. And then you write a description of what that feature is, and you just let ChatGPT do all of the magic. You can regenerate the response if it's not what you want. And you can ask it to generate more responses if you just need more, almost like brainstorming, just to make sure that you have included all of the required user stories into that specific feature for it to be completed. So in a way, this could also be a really great brainstorming tool for you as well. I hope you are just as mind-blown as I am by these tools. I hope you enjoyed today's lecture. Make sure you Subscribe to our channel for more great content like today's, and I'll see you in the next video. Bye for now.