How I Use AI to Brainstorm Product Features (No Meetings, Just Results)
Lately, I’ve found myself turning to ChatGPT for all kinds of things.
“I have three potatoes, a carrot, and two cloves of garlic in my fridge. What can I make for dinner?” (Hint: not much.)
“I have an idea for a startup with these criteria. Can we build on this?” (Spoiler, yes. It can.)
The reason for this is not because I intend to take the first suggestion ChatGPT comes up with and call it a day. Rather, as an external processor, I can’t begin the problem-solving process without getting my ideas out into the world—preferably into the hands of a willing participant.
A brainstorm with only one brain is just a drizzle.
But in the era of remote work, it’s not quite as easy to find someone to workshop ideas with at the drop of a hat. By the time I’ve decided who I should chat with, check their calendar, and schedule a call for next Tuesday…the idea I wanted to discuss is a distant memory.
To combat this problem, I’ve often turned to tools like Notion and Miro to assist me in bringing definition to my thoughts all by myself. I’ve become quite proficient at this, though I still crave a sparring partner to help nudge my thinking along, even if only with probing questions and thoughtful encouragement.
Little did I know that ChatGPT would prove to be an acceptable stand-in. Almost as compelling as the LLM’s conversational capabilities, ChatGPT is never in another meeting, can operate in any time zone, and never gets frustrated when I ask the same question three different ways to prompt a slightly different response.
It wasn’t long before I started using ChatGPT to help me workshop product functionality. Before I knew it, I had identified a few tips and tricks to get the most out of this tool.
How to Use AI to Brainstorm Product Features
To illustrate my AI-powered brainstorming technique, I’ve created a sample conversation with ChatGPT about building features for a fictional health and fitness application.
To prevent the LLM from pulling data from all of our previous conversations, I conducted this conversation while logged out.
It's worth noting that ChatGPT is my tool of choice, but the same methodology should translate well to any similar LLM.
Step 1: Provide the Context
The key to a productive ChatGPT interaction is the initial prompt. The more context you can provide in your first message, the more fruitful your interaction will be.
In this example, I wrote a scenario for a product manager attempting to develop feature ideas for a SaaS fitness tool which intends to help users customize their workout routine.
The LLM was easily able, based on a limited set of information, to craft a long list of feature ideas for the product I described. This process isn’t about taking this list and running with it. It’s about getting to the first stage of brainstorming faster than I would have been able to on my own.
It can be a thought partner. It can simulate a hard conversation. What’s the most important thing you should do next? The answers to that are really amazing.
Step 2: Going Deeper
Now that I’ve got a list to work with, it’s time to evaluate and workshop the ideas that I’m interested in. While reading through the list, one idea in particular piqued my interest.
“Smart Scheduling and Availability Management” appeals to me as a potentially unique offering in the fitness world today.
A quick Google search confirms that products offering this service in the fitness category are available but fairly scarce. It seemed like there was an opportunity to add something new to this problem space. Plus, as a busy working mom of a 14-month-old, my dream of working out at the same time every day is a distant memory.
I ask the ChatGPT to further drill down into that idea to see what an expanded offering might look like.
I was impressed by the expanded functionality that the LLM sent back. My next step was to ask ChatGPT to map a user journey for me, including the functionality above. I provided a few parameters to refine the response.
It’s important to keep in mind that ChatGPT pulls data from the web, so it’s most likely pulling ideas from products that already have a presence online. However, it’s still a great way to mine ideas and kickstart the brainstorming process.
Another option, at this stage in the ideation process, would be to ask ChatGPT to examine the items on the market today and identify a market gap. This will prompt the LLM to analyze the items that already have a presence online and make suggestions about less saturated ideas. Of course, all suggestions require human verification, but it can be a great way to narrow down the options.
Step 3: Refine the Idea
At this stage, we’ve chosen the idea we’re going to explore further, and we’ve verified with some high-level research that the idea is somewhat unique. Now, it’s time to workshop this idea into a tangible user journey.
In this step, instead of providing a scenario prompt to ChatGPT, I provided a set of instructions. Specifically, I asked ChatGPT to include “any or all” of the features described above to prevent it from trying to include each piece of functionality regardless of whether it could logically be placed into the user journey.
Furthermore, I asked ChatGPT to indicate the difference between user-driven actions and application-driven actions in its writeup. Lastly, I asked the LLM to map the journey from login to logout.
In addition to following my instructions by separating user actions and application actions, ChatGPT even went as far as to create a high-level user persona around which to design the user journey, which I did not ask it to do.
The readability of the user journey was somewhat inhibited by the limitations of ChatGPT’s output capabilities (i.e., in the free version, ChatGPT can only respond using text, as opposed to diagrams and other visual communication tools).
However, I felt that this was partially mitigated by ChatGPT’s inclusion of a summary at the end of this segment which provided a high-level overview of all user actions versus machine actions.
A few notable omissions stood out after reading through the user journey. The first was the absence of a process for integrating the user’s personal calendar with the application calendar, which would provide the data for the dynamic scheduling feature.
The second was the absence of an alerting feature to remind the user when they had a workout coming up. I documented these adjustments and added them to the conversation. Each time, the LLM sent back updated user journeys that accounted for the changes I requested.
Step 4: From Ideas to Execution
The final step in this experiment was to turn these feature ideas and user journeys into a tangible, high-level “Now, Next, Later” roadmap that could theoretically be executed upon. Here’s the prompt I used to start that process.
ChatGPT returned a comprehensive list of requirements broken down into user stories and acceptance criteria categorized into "Now, Next, and Later" roadmap items.
I did not provide definitions for what constitutes a "Now, Next, or Later" item in my roadmap. In the absence of these definitions, ChatGPT determined its own based on explanations it found on the web. I was satisfied with these definitions, so I didn’t request adjustments.
It is, however, important to call out that the more parameters you provide ChatGPT, the more tailored to your preferences your output will be.
Accelerate your capacity and creativity
In conclusion, ChatGPT and other LLMs can provide a valuable sounding board for workshopping your product ideas.
Personally, I find the first stage of product ideation to be the most challenging and where I require the most external feedback. This is where I prefer to leverage ChatGPT to assist me in getting over the hump of early-stage brainstorming. I also will often use it to provide some high-level validation around the uniqueness or positioning of a particular idea.
Should you choose to, however, ChatGPT and other LLMs can take you far enough in the process to arrive at a high-level roadmap that can be easily built into a more tangible plan. When used properly, these tools have the potential to dramatically accelerate the capacity and creativity of any product team.
Don't forget to subscribe to our newsletter for more product management resources and guides, plus the latest podcasts, interviews, and other insights from industry leaders and experts.
