Before building something, I like to check if someone else has already built it. So I spent a day doing a deep competitive sweep of every tool that lets you query multiple AI models from a single interface.
I found over twenty. The space is crowded, well-funded, and growing fast. Poe, ChatHub, TypingMind, ChatPlayground, Perplexity's Model Council, Together.ai's Mixture of Agents, and at least fifteen more.
And not a single one lets you highlight or strikethrough the AI's response.
I had a moment last week that probably sounds familiar.
I was deep in a conversation with an AI assistant — five turns in, discussing something strategic and nuanced. The AI gave me a long, detailed response. Three paragraphs were exactly what I needed. One paragraph was completely off base. And one sentence buried in the middle was genuinely brilliant — the kind of insight I wanted to build the entire next turn around.
So what did I do?
I typed a message. A long one. "I agree with your first point about X, but the part about Y is wrong because... and that sentence about Z — that's exactly right, let's go deeper on that, but ignore the stuff about W..."
By the time I finished typing, I'd spent more time writing my feedback than reading the original response. And I knew that half my nuance would get lost anyway — the AI would latch onto some parts of my feedback and miss others, because I was trying to communicate a complex, multi-layered reaction through a single text box.
This happens every day, to millions of people, across every AI chat interface in the world. And nobody seems to be working on it.