Monogram AI

The on-the-fly generated UI per response is the actual bet here — most "AI + interface" attempts just re-skin a chat bubble, this genuinely renders a fresh layout each time. Day-one design question: is the generated interface deterministic for a given query (so re-asking "what should I watch tonight" gives me a stable, recognizable view), or does it re-roll a new layout every time and make the app feel unpredictable? And can I hand a generated interface to someone else — send a friend the interactive trip plan — or is it locked to my session?

the "generates a whole UI on the fly" part makes me wonder about accessibility. a fixed app UI can be tested once with VoiceOver and you know it works. if every response is a freshly generated interface, how do you make sure that stays usable for someone relying on a screen reader, or is that just not solved yet at this stage

Finally tried this and the recipe one genuinely surprised me, it pulled up a full ingredient checklist with step navigation instead of a paragraph. Felt like a real shift from the usual chat walls.

How does Monogram handle the UI generation when it needs real-time data like live sports scores or current flight prices — is it pulling from specific APIs or is everything happening client-side?

Is it possible to do a particle animation when switching between different layouts, so that it all looks fluid ? That would be cool

@edouardtabet Interesting idea, and pretty convincing demo! Some questions though: isn't generating interface everytime token/hardware consuming ? what kind of LLM is behind and is that customizable ?

How does it decide which type of interface to generate for a given question, and is there any way to customize or save the layouts it creates?

How does it handle follow up questions, like if I want to tweak one of the suggested options it just generated? Does the whole interface rebuild or can it edit just that part?

I tried the app and really loved the idea! AI is rendering widgets/tools with each response and creates a unique UI for each output.But I have a question -- right now I see that you don't store previous conversations (or at least I couldn't locate it). Are we going to be able to leave certain chats open like we do on Safari tabs. I may want to be able to switch between tabs and continue with a previous chat I had.

Are these interactive interfaces shares like ChatGPT chats?

Hey everyone. We believe AI deserves a better interface than chat, this is our first take! We would love to get your feedback and your thoughts on the future of UI for AI.