Daniela Amodei, president and co-founder of Anthropic, stated that the current intensive use of AI by enterprises is still in its early stages. As model capabilities continue to improve, the value of AI in workflows will gradually become clearer, rather than remaining at the stage of "using it for the sake of using it."

Discussing the phenomenon of high-frequency use of AI

At the Bloomberg Tech conference in San Francisco, Amodei addressed the recent Silicon Valley buzz surrounding "tokenmaxxing." This term typically refers to developers using AI tools as much as possible, consuming large amounts of tokens and incurring high bills, with short-term commercial returns not always clear.

She stated that the model's capabilities have improved significantly over the past two years, but this is not the end. She anticipates that the range of tasks the model can accomplish will continue to expand over the next two to four years, or even six to eight years. Enterprises will also find new ways to use it in practice and gradually develop more mature collaborative experiences.

Claude Code sparks heated discussion

This increased discussion is related to the rapid proliferation of AI programming tools. Anthropic's Claude Code is considered one of the representative products. Compared to traditional chatbot interfaces, these tools typically consume more tokens, especially as agent-based AI can continuously and autonomously handle tasks, further increasing the number of calls.

The increased demand for AI programming, driven by token-based pricing for some Anthropic and OpenAI products, has boosted related revenue. This has led to more internal discussions about whether the return on investment is justified.

• A token is a unit of measurement for AI inputs and outputs.
• Agent-based AI can improve the throughput of continuous tasks.
• The token-based billing model amplifies changes in usage costs.

The company does not have a ranking system.

Amodei also mentioned that some employees are adapting to the new work environment. Some companies require employees to use AI tools and even track their usage. She believes that in the future, AI is more likely to be naturally integrated into daily work rather than being promoted through leaderboards or mandatory performance evaluations.

She stated that Anthropic observes the team's overall usage of Claude products, but the company does not have an internal ranking system for "must use AI" or "must use Claude".

Previously, Amazon executive Dave Treadwell had also warned employees not to use AI for the sake of using AI. According to The Information, an internal Meta leaderboard related to Claude usage has also been removed by its creator.