Summary

I would focus on:

Onboarding: Explain to people all the ways in which they can use the tool. From my own experience and cursory research it seems that users are mostly limited by their own imagination. Most people don’t think of things to do with GPT
Answer UI: Once you are using it, expand the ways in which you can use it (similar to Onboarding above) but also make it easier to fall into a quick rhythm with the model where you are trying things.

Onboarding

🛑 The app did little to tell me how to use it (four example prompts that are not front and center).

🛑 The main action (input field) was already hidden by “NEW” notices in different parts of the app. And color guides me away from the input field

🛑 If I run out of ideas of what to try in a thread, the app doesn’t suggest ways to continue. It is easy to fall off the train.

Answer UI

The Q&A UI is clear in how you interact with it but doesn’t help you fall into a quick rhythm with it which is key to sustained usage.

🛑 As the answer streams back, the vertical position of the answer box moves, making it very hard to use. I’d pin the answer to the top while the answer streams.

🛑 I highlight what interests me of the answer and nothing happens. It’s a lost opportunity for quick iteration with the model. Some ideas for a tooltip:

“Explain this”
“Ask a follow up about this”: copies the text into the input field

🛑 When I realize I didn’t prompt it right, I have to go copy and then paste the original question into the input box. I should be able to CLTR+UP quickly into my latest message to try again.

Ideally, it would teach me about it if it seems like I need it.

🛑 If I run out of ideas on what to do next, the UI doesn’t help me keep going. Most other UIs suggest several follow ups with each response.