How we "revolutionized" AI user experience

While Big Tech was chasing smarter models, they forgot something obvious: humans.

We tried all the big AI tools. We paid for them. We hit their limits. And we realized something ridiculous: nobody was fixing the basics. So we started from scratch — not by making AI "smarter," but by making it actually usable.

Delete message button

So you can clean your context.

Sounds obvious, right? But try deleting a message in ChatGPT. You can't. You can only edit it, which triggers a ripple effect where your entire chat history gets rewritten. Chaos.

Claude? Perplexity? Same story. No delete.

In loqus.ai, you can just… delete the message.

And boom, your AI stops hallucinating about that offhand comment you made 40 messages ago. Clean slate, no drama.

Copy tables like a normal person

We were shocked at how painful this still is.

╯°□°）╯︵ ┻━┻

Until just recently, exporting a table in ChatGPT was a mess. We had to switch to "data analyst GPT" and ask it to generate a CSV file. Sure, you could try selecting the table with your mouse — but it was clunky, especially with wider tables. It stayed that way for years… and only got properly fixed a few days ago.

Claude lets you highlight too, but there's no copy button.

Perplexity? Download as CSV (though it is nice to have), but forget about just copying to clipboard.

In loqus.ai you can do all:

Select with cursor
Copy via button
Download as CSV

No hacks, no friction — just the thing you expected to be there in the first place.

┬─┬ノ( º _ ºノ)

Copy answers as clean markdown

OpenAI sort of has it. Until you paste it and your doc is full of weird characters. Also, good luck with broken URL formatting.

Perplexity? Same issue — they fixed it while exporting files, but with copy button the links aren't clickable.

In loqus.ai, markdown just works.

Pasting content into Notion, GitHub, or a doc? It looks the way you want it to. Clean code blocks, proper URLs, readable formatting. The boring but necessary stuff — done right.

Search the web

We've spent 20 years perfecting search — and then AI came along and made it clumsy again.

Have you noticed that In ChatGPT, you can analyze files or search the web. Not both. Trying searching something from the screenshot you've just made requires at least two steps.

Have you also noticed how hard it is to ask for additional search results when you definitely know there are more and you explicitly point it out?

Claude's web search still isn't available outside the US. Who needs the rest of the world, right?

Perplexity is kinda associated with AI search, right? It already has two different built-in search modes but for some reason builds an external browser to handle agentic search. But why make it harder?

Btw, if you want to know more how cool AI search is, check our article about it.

We do it differently:

You chat, you upload files, you search, you get results — all in one place.

Remember when search was simple? So do we. That's why we built it straight into your chat window.

Also, we have

✨ Top AI models from OpenAI, Anthropic, Perplexity, DeepSeek, Qwen and others

✨ Task-specific custom agents

✨ Voice input and voice command, so you can interact with your input message before sending it

✨ File handling that supports a majority of text and image file types

What we don't have

🧢 A merch shop

🕸 A separate web browser

🤯 A phone number to call your AI

If you are ok with that, try loqus.ai today. Then tell us:

Did we actually do a good job?