As a user, I’m not a big fan of apps that on the 1st start explain all their functions. I physically hate clicking through never-ending "next" buttons on their tutorials, pointing me to next important place on the screen. There has to be a better way. When you open the app, you should know what to expect. Like walking into the supermarket: you open the doors, you pick up a basket, you walk to the shelves, select your favourite products, and put them into the basket. You repeat the last step multiple times until you eventually go to the till, where you pay. Perfectly simple and obvious.
But then, what makes you go to a specific shop? What actually makes you select one over another? At the end of the day, they are all the same; they just slightly differentiate on the User Experience (UX). Product selection, overall care, and having physical staff are differentiators. Still, as Peter Thiel famously said, "Competition is for losers.” You ultimately want to create a product that doesn’t exist.
So you create one. A new, unique proposition where you are the settler. Congrats, you’ve designed your unique cold fusion toaster.
How the hell are you now supposed to create an experience of walking into the shop for something this radically new? Something that people have no neural pathways created for? Do you start from, "It’s just a toaster, but cold"? Do you drive users through the sci-fi cold fusion dream? Is there any widely known mental anchor you can use?
You keep trying new ideas. You test them. You remove them. You rinse and repeat.
You flirt with tutorials, even though you know that they are just an ersatz of proper onboarding:
You add contextual hints:
You rediscover the ancient and battle-proven pattern of Wizards:
Jokes aside, when you squint a bit, the wizard paradigm drives the majority of the services under the gov.uk domain. Step-by-step is a deeply researched (maybe the best-researched user flow) and proven design paradigm that works for many. It may be boring for experts, but it’s delightful for occasional users. On the surface, you have a page that asks you a single question and provides enough context so that you, the user, can give an informed answer. The next page does the same, until the service has all the needed information to process the transaction.
The pattern has been polished further by the GOV.UK Design System with Step by Step Navigation.
The beauty of this pattern is that it leaves you enough space to teach the user about cold fusion and toasting. An advanced user will just press “next” and whizz through the experience. It leaves the option to the user. As a designer, you have a whole page to experiment with advanced controls.
Application
The big problem with Flash Notes is that it’s universal. You can use it to memorise anything1. You can learn any language, and in any language2. On top of it, you can ask the machine to generate the flashcards for you. You specify the subject of study by just naming the note, and the generation picks it up3. It’s like KitchenAid, but you don’t have a shop assistant helping you to select the right add-on. On the plus side you have all the add-ons already built in, but you are on your own to discover it.
How do you explain this creative freedom? Cookbooks exist for a reason. Unless you are an expert and enjoy experimentation you won’t know how to make an omelette just from opening the box of eggs. What’s worse, if you’ve never had an omelette before, how would you make one? However the omelette possibility is in the eggs.
Creating product is an art of limiting choice so it’s helpful for the user. Otherwise we would be just shipping BASIC and Excel and a language model (GPT) on the PCs.
Ideally, you guide user through examples and drop-downs. You hint the possibility. And that’s where step-by-step shines. You have the user's attention and screen space to give these examples when they are mostly needed - for Flash Notes it’s the moment the user is creating a new note.
Try the new Flash Notes Wizard interface experiment and let me know what you think.
I find the process of machine creating a study material… magical
Anything as long as the prompt and answer are textual. Textual means readable. Readable means you can learn while you walk.
Currently limited to languages natively supported by the operating system for text-to-speech, but let us know if you need more.
Quality Assurance of LLM outputs using G-Eval is amazing and a massive pain - separate technical blog post candidate.