Most companies have ideas for generative AI projects but struggle to get them going. They assume a big initial investment in data, skills, or infrastructure is necessary to get started. But in almost all situations it is better to start with rapid prototypes before making any such investments.

Prototyping in generative AI is different from prototyping in web or mobile apps. You don’t want to spend time building wireframes, mockups, or user stories. Instead, you want to quickly answer three questions about your proposed AI product:

1. Are the models smart enough?
2. How much data investment is required?
3. Will the solution be useful?

You can begin answering these questions very quickly in generative AI. The primary steps are:

Step 1: Prepare the data.

Step 2: Process the data with prompts.

Step 3: Qualitatively evaluate the results.

You should be able to get through these steps in days or even hours. If you’ve been working on your generative AI project for more than 2 weeks and you don’t have early answers, you’re almost definitely doing something wrong. 

‍I also share how we’re doing rapid prototyping with project Ground Crew, our example application for this step-by-step mini series.