As a product manager, I strive to utilize tools with the highest efficiency. During my MSc in AI for Business, I had the opportunity to learn about how Gen AI works. One of the most crucial concepts is tokenization. In this blog, I will break down exactly how to get the most out of your favourite Gen AI tool.
In my previous post, I explained three ways I use Gen AI. As a product manager, I strive to utilize tools with the highest efficiency. During my MSc in AI for Business, I had the opportunity to learn about how Gen AI works. One of the most crucial concepts is tokenization. In simple terms, tokenization is the concept of breaking down a prompt into smaller words so that the GenAI can understand what the user is asking for. You can provide your GenAI tools with more accurate prompts to obtain the correct outputs. In this blog, I will break down exactly how to get the most out of your favourite Gen AI tool.
One of the first steps to get the best out of your GenAI tool is to understand its capabilities. Each platform, whether it's Gemini, ChatGPT, or Claude, has different versions that are best suited for specific tasks. Ask the tool if it is capable of completing the task you aim to finish or if a better version is available. When I use GenAI to create images, I ask it what types of images it can generate and select a style that suits my needs. Adding the exact prompt that aligns with the capabilities of your GenAI tool will give you more accurate results.
When building rapid prototypes, your GenAI must understand the end users of the product. Clearly mentioning your target market or outlining your customer personas will help your GenAI build prototypes that can seem helpful for your customers. The reasoning is simple: GenAI is capable of understanding user behaviours based on the training it has undergone. So, instead of worrying about the colours that best suit your prototype, let your GenAI do the heavy lifting. Here is a simple example of what a prompt I would use.
"Hey GenAI, I want to build a rapid prototype for helping restaurant managers who are typically between the ages of 30 and 55 with staff scheduling".
For your Gen AI to provide accurate results, you need to clearly outline the actions your users can perform on your prototype. If you are unfamiliar with user stories, here is a quick formula to follow. First, outline the types of users of your product, then describe the goal they are trying to achieve and state what they are happy with if the user can reach their goal. By clearly providing detailed user stories, you help your GenAI organize its thoughts and actions.
Here are some user stories for my prototype, designed for a restaurant manager to manage their shifts more effectively. The only user of this product will be a restaurant manager.
Once you have identified the user stories and your target market, clearly state the tools you intend to use. To do this, you need to first understand some basic software development terms. GenAI tends to produce a lot of noise if allowed to be creative; it can also hallucinate and produce inaccurate results, which can be very demotivating. Here are some quick tips: if you are building web apps, clearly mention the framework you prefer, the behaviour on smaller screens, whether you want to use a database, and you can even say the file name. Now, to add to my above prototype of a restaurant schedule manager, here is an example prompt:
"Build a single-page web app using HTML, CSS, and JS that is responsive and uses Local Storage for data persistence."
Of course, we all have our favourite GenAI tool, but it helps to know which GenAI is best suited for what you are building. As mentioned earlier, be sure to ask your GenAI what it excels at and pick a model that aligns with your needs. Using the examples above, I have built three different versions of the schedule management app using the following GenAI tools:
With the help of GenAI, you can build many prototypes and see if they can help you solve problems in your daily life. To obtain accurate results, you can conversationally interact with the GenAI, much like having a conversation with it. The process of engineering your prompts to ensure you get more accurate results is called prompt engineering, and by doing this, you can use your GenAI very effectively. Always keep in mind that each GenAI is trained for specific tasks, and your specificity can help the GenAI provide you with more accurate results. Here is the final prompt I used to build the schedule management software for restaurants:
Try these techniques with your next Gen AI prototype and let me know how it goes!
Get weekly insights on AI prototyping, product management, and building better products with AI. Join 1000+ product leaders who are leveraging AI effectively.