ChatGPT Vision — Five Real Use Cases for Product Managers (Part Two)
Using Vision Capabilities to Support Rapid Prototyping
In part one, we learned how to access GPT-Vision and looked at our first use case: transcribe and transform notes. A key idea I have begun to tease is the concept of “task-oriented conversations”. For transcribing and transforming notes this means taking a single starting point (like sticky notes from a workshop) and then building upon that starting point with an in-depth conversation that becomes richer and more detailed over time. In our last article, this meant taking a workshop output and turning it into a task list, then a two-week plan, and then a calendar schedule.
Although AI tools are great at generalisation—and there are plenty of use cases for general-purpose chatbot conversations—we will stick with this idea of a task-oriented conversation. I think ChatGPT is great for helping product managers to prototype rapidly, and the release of Vision means you can take a single idea and rapidly expand it into text, code, and images. If you want to jump straight to the practical examples examples read on, but if you haven’t read my recent article about prompting I would highly recommend starting there and coming back as it will provide extra context for how and why I write my prompts.
Rapid Prototyping with ChatGPT Vision
You can employ rapid prototyping to help you test and validate ideas quickly throughout the product development process. We are going to look at examples that leverage Vision across:
Discovery
Design
Development
Post-launch
Discovery
Discovery is one of the most important stages in any product development process. During the discovery phase, the product manager works with a cross-functional team to define the vision and strategy of their product, ensuring that everyone understands what problems are being solved.
Rapid prototyping inside the discovery phase can let you test and validate hypotheses quickly, and ChatGPT can be an excellent companion for this. As I mentioned in my prompt engineering post, treat ChatGPT as a team member and bring them into your workflow. Use them as a sounding board when doing individual work, or “invite” them to a workshop and have participants interact with ChatGPT via you.
As you begin your discovery process, you could take some screenshots of competitors’ products and ask ChatGPT to help you ideate to inform your first prototypes:
Here’s an example using “FutureSaver”, an idea ChatGPT came up with in my previous post.
ChatGPT completed an analysis of each app screen one by one and then came up with some recommendations to inform how we design and prototype FutureSaver:
Fun Workshop Idea:
At this stage, a fun workshop idea is to get your entire team onto a shared space like Miro, Figjam, or a real whiteboard. Using only basic shapes and text, get all participants to mock up what they think the end product should look like (blue sky thinking, take away all constraints). Once they are done, take a screenshot or a photo. Tell ChatGPT to act like a judge and run a virtual competition where ChatGPT judges each submission.
I cut the chat for brevity, but this kind of “augmented” workshop is an excellent use case for bringing AI into your product and design activities.
Design
Earlier, ChatGPT pulled out the key elements to draw upon from our competitors. We can then directly take the recommendations from ChatGPT and input them into a Figma plugin called WireGen or have a collaborative session with our designer or stakeholders.
Once we have a design, we can use ChatGPT to iterate and improve. By uploading our wireframes (either AI or human-generated) back into Vision, we can get immediate feedback to improve on future versions.
Rinse and repeat. Receive AI feedback, implement changes, ask for feedback again, and so on. While you are working with designs, you can also use ChatGPT to assume a persona to test for usability and accessibility.
I uploaded a high-fidelity app mockup into Vision and received five different pieces of feedback. Here is one example where the mockup was analysed from the perspective of someone with visual impairment:
Does this replace real user testing? NO! But you can solicit feedback in seconds and rapidly prototype in a short amount of time.
Development
Once you are building your product, the prototyping doesn’t stop.
Provide ChatGPT with a list of acceptance criteria for a screen and then upload a work in progress for a quick sense check:
Or, maybe you don’t even have your acceptance criteria yet. No worries, get ChatGPT to turn an image of your wireframes into a set of user stories and AC’s.
Post-Launch
Once your product is live, there is no reason why you couldn’t use some of these techniques again as part of your continuous improvement efforts. However, I will throw in one extra idea: marketing.
Bring all of your marketing and advertising materials into ChatGPT and incorporate them into the conversation. Get feedback, ideas, and rapidly prototype before going live into the marketplace.
Conclusion
Hopefully these examples have shown some of the many ways you can use ChatGPT and Vision to rapidly ideate and prototype across the entire product development lifecycle. As with most gen AI applications, your potential is only limited by your imagination. Start by introducing Vision into some of your workflows and see how well it can complement you and your team.
I deliberately haven’t talked about using Vision to write code; we’ll take a look at this use case in part three. See you there!
I am publishing weekly articles on “using AI” and “building AI” products here and on my Medium. If you subscribe here you will get exclusive access to all articles one week early, for free. And if you think any of your friends or colleagues would benefit from my content I would love it if you could share this post.
Thanks for reading AI Product Management! Subscribe for free to receive new posts and support my work.