No more BuzzFeed tone: Gemini cleans up Claude's AI answers

Artificial intelligence sometimes has a tone that's a bit too, shall we say, "lively." It talks like it's trying to sell you something, or worse, like it's hosting a kitten ranking show. A nightmare if you're looking for seriousness.

Many AI models, including Claude, tend to use overly enthusiastic language, full of exclamations and clichés. It's the so-called "BuzzFeed voice," a blend of clickbait and superficiality that gets tiring for anyone. Who wants a bot acting like an influencer?

This tendency makes it hard to use AI for serious or professional content. You certainly don't want your business report to sound like "7 Epic Meeting Secrets Your Boss Doesn't Want You to Know." Adnan Akil developed Nobuzz to counteract this trend in AI-generated language.

The trick is simple yet brilliant: you make one AI correct another AI. Adnan Akil created Nobuzz, a "skill" for Claude that essentially takes Claude's "overly cheerful" answers and feeds them to Gemini for a refresh.

Gemini, Google's model, receives Claude's output with a specific task: rewrite it, removing any trace of that viral listicle tone. Think of Gemini as the sarcastic, unforgiving proofreader who doesn't tolerate an extra phrase. Not a bad idea, right?

Technically, Nobuzz acts as a filter. When you ask Claude something using the /debuzz skill, its answer is processed by Gemini. The final result is a text that's drier, more professional, and, let's be honest, significantly less embarrassing. The /debuzz skill in Nobuzz routes Claude's responses through Gemini for stylistic revision.