Digital Thinking Will Destroy Your AI Products' Potential. Here's A New Approach.
How long do you think it’ll be before LinkedIn offers a ChatGPT-powered assistant that helps students to prepare for job interviews? This is one of the hundreds of use cases sitting just under the surface for our new crop of large language models. It will only get bigger and wilder over the next 24 months.
The next generation of GPT models will incorporate text, images, and video. Those models are right around the corner and will open up an even larger landscape of use cases and potential applications. We’re not discussing some of the pitfalls and trouble you’ll encounter while implementing them.
There’s a massive difference between having a functional model and being ready to deliver an AI-supported product. The model is just the beginning. Other considerations revolve around the user and their potential applications rather than the model itself.
Always Start With The Workflow
The first place I begin the planning process is by evaluating the workflow itself. What is my intended user or customer going to do with this product? Most AI-supported products start backward. The technology enables a potential application, and someone develops the AI product because it can be built. Users don’t typically adopt this paradigm.
I must understand and document what users are doing right now. Job interview preparation is a complex and messy process. One option is to hire someone who coaches you through the preparation process. There are classes and boot camps that prepare people for job interviews, especially in the technology field.
Gumroad and the like are filled with thousands of cheat sheets and interview prep forms. Some promise to deliver the questions and answers to prepare you for an interview at specific companies. Search for Amazon Software Development Engineer L4 Interview, and you’ll get several options.
Mapping the workflow requires documenting how each option is used. This is a deep dive into job seekers’ steps while discovering, using, and interviewing with these products. Most people stop at the using part, but there’s a critical difference involved in AI products that gets missed.