The best non-chat AI interfaces I’ve seen
The next generation of human-AI interfaces
I’m obsessed with UX, especially when it comes to AI. Most AI products today default to chat, likely because the big labs all lead with chatbot interfaces. But chat is just one way to interact with intelligence. Lately, I’ve come across a few products that offer a glimpse of what post-chat UX might look like.
Here are some of the most interesting patterns I’ve seen:
1. Async co-creation: human and AI take turns
Granola, Cursor’s predictive editing (tab), Cove.ai
In Granola, human and AI split the task of note-taking: the user captures “what I think,” while the AI fills in “what they said,” based on the transcript. It’s a clean division of labor, and it works.
Cursor’s predictive tab and Cove’s spreadsheet fill follow a similar rhythm: the human starts, the AI continues. You don’t prompt—it picks up where you leave off.
These interfaces feel more like jazz than ping-pong. Less reactive, more collaborative.
2. Personalized content engines: recommender meets generator
NotebookLM, Particle, Google Daily Listen
Imagine if Spotify didn’t just recommend songs—it composed them just for you. That’s the idea behind this new class of products.
NotebookLM generates a podcast based on content you upload. Google Daily Listen goes deeper, using your search history to create a daily news podcast tailored to your interests. Particle takes it further: a personalized news feed written entirely by AI, with tunable tone and style.
The future isn’t just personalized discovery—it’s personalized creation.
3. Micro-feedback interfaces: intent signaled by a button
In Quill, a “Highlight” button lets you flag moments during a meeting. Later, the AI writes up what was said before and after that point. Snipd applies the same logic to podcasts: tap “Create Snip,” and the AI generates a summary card from that timestamp.
These interfaces make AI feel ambient—less like a tool, more like a silent partner that listens and helps when you nudge it.
Final Thought
These products suggest a future where chat is just one of many ways to work with AI. The best interfaces are invisible until they’re not, surfacing only when you need them, and vanishing when you don’t.
We’re only just beginning to explore what AI-native UX can be.
Cove AI is my fav. In fact, I built my fitness program on it: https://aimaker.substack.com/p/how-i-built-my-entire-fitness-program
I hope soon we'll move away from chat interface when interacting with AI.
Love this and all the examples. It’s a true exemplification of “disruptive technology” which is defined by being disruptive in value and seamless in experience!