Major AI companies are deliberately making their chatbots less conversational and more blunt to cut costs, with one plugin reducing token usage by 65% by removing pleasantries and unnecessary prose. The shift exposes a fundamental tension in the AI industry: after years of marketing artificial intelligence as remarkably human-like, companies are now prioritizing the bottom line over personality.

What's Driving This Cost-Cutting Trend in AI?

The AI industry is facing a profitability crisis. Many of the most popular AI firms have recently shifted from flat-rate subscription models to per-use pricing, reflecting the prohibitively expensive upkeep of these services. This has forced major companies that embraced AI to become more conservative with their spending.

The financial pressure is real and widespread. Earlier this year, OpenAI shuttered Sora, its video generation tool, which was bleeding approximately one million dollars per day. The shutdown complicated what was otherwise a lucrative partnership with Disney.

In response to these mounting costs, companies have begun experimenting with unconventional solutions. A new plugin called "Caveman" is making the rounds among organizations looking to reduce their token usage, the computational units that determine how much an AI service costs to operate.

How Does the Caveman Plugin Work?

The Caveman plugin instructs large language models (LLMs), such as Claude, to strip away the conversational elements that make AI responses feel natural and human-like. Instead of friendly advice with transitions and pleasantries, the plugin produces curt, direct responses.

Julius Brussee, the creator of Caveman, developed the plugin in early April after noticing that much of his token spending was going toward unnecessary elements. He explained his motivation and results in detail.

"I made Caveman back in early April because I was using Claude Code heavily and noticed a lot of my token spend was going to unnecessary prose, pleasantries, hedging, transitions, and chatty language that does not really matter," said Julius Brussee, creator of the Caveman plugin.

Julius Brussee, Creator of Caveman Plugin

The results have been striking. By reducing what Brussee describes as the "IQ" of AI responses, his token usage dropped by approximately 65 percent. This dramatic reduction demonstrates how much computational overhead goes into making AI sound friendly and conversational.

Ways Companies Are Adapting to AI Cost Pressures

• Reducing Conversational Elements: Stripping away pleasantries, hedging language, and transitions that add to token usage without adding substantive value to responses.
• Shifting Subscription Models: Moving from flat-rate pricing to per-use models that make companies more aware of and conservative with their AI spending.
• Shuttering Unprofitable Services: Discontinuing AI products like Sora that cannot achieve profitability despite strong market demand or strategic partnerships.
• Adopting Cost-Reduction Plugins: Implementing third-party tools designed specifically to minimize token consumption across AI operations.

What Does This Mean for AI's Future Direction?

The irony is difficult to ignore. The AI industry has spent years marketing these systems as remarkably intelligent and human-like, emphasizing their conversational abilities and emotional resonance. Pop culture has reinforced this narrative, with companies going to "absurdist lengths to anthropomorphize language learning models," as one observer noted.

Now that profitability actually matters, the industry is rapidly reversing course. The shift from human-like intelligence to blunt efficiency represents a fundamental recalibration of what AI companies believe their customers actually need. It suggests that the emphasis on personality was partly a marketing choice rather than a technical necessity.

None of the major AI firms have found a way to be naturally profitable, which means cost-cutting measures like Caveman are likely to become more common. As token costs continue to rise and competition intensifies, expect more AI companies to prioritize efficiency over conversational warmth. The chatbots of the future may be less charming, but they could be significantly cheaper to operate.