The video criticizes Anthropic for implementing unfair and convoluted billing practices in their Claude Code product, including charging users based on system prompts and even filenames in git commit messages, which penalizes users regardless of actual usage. The speaker expresses frustration over these policies, arguing they alienate the engineering community, damage user trust, and call for Anthropic to fundamentally rethink its approach to regain credibility.
The video expresses strong frustration with Anthropic, criticizing the company for repeatedly making questionable decisions that negatively impact users. The speaker highlights how Anthropic has implemented billing practices that change depending on the system prompt used in their Claude Code product. This approach was apparently designed to discourage users from integrating Claude Code with third-party tools like OpenClaw, which Anthropic does not officially support. While the speaker acknowledges the financial reasoning behind this—citing the high cost of inference requests—they argue that it is unreasonable to penalize users simply for using different tools or system prompts, especially when their actual usage remains within their subscription limits.
The criticism deepens as the speaker reveals that Anthropic has escalated its billing policies to an even more absurd level. Now, users are being charged based on the files present in their codebase, and more specifically, on the filenames included in their git commit messages. This means that even if a user has not actively used Claude Code, they can still be billed if certain keywords appear in their commit history. The speaker shares a concrete example of a user who was charged $200 despite not consuming any usage, simply because their commit message contained the phrase “Hermes MD.” This practice is portrayed as both unfair and nonsensical.
Throughout the video, the speaker conveys a sense of exhaustion with repeatedly addressing Anthropic’s missteps but feels compelled to continue due to the severity and absurdity of the issues. They describe Anthropic’s actions as embarrassing and indicative of a deeper incompetence within the company. The tone is one of incredulity and disappointment, emphasizing that these problems are not minor glitches but fundamental flaws in how Anthropic manages its products and treats its engineering user base.
The speaker also touches on the broader implications of Anthropic’s approach, suggesting that the company’s disdain for engineers is evident in these policies. By penalizing users for factors unrelated to actual usage, Anthropic appears to be alienating the very community that relies on its tools. This disconnect between the company and its users is framed as a significant problem, undermining trust and usability. The speaker’s critique implies that Anthropic’s current trajectory could harm its reputation and user loyalty if not addressed.
In conclusion, the video serves as a pointed call-out of Anthropic’s billing practices and overall handling of its Claude Code product. The speaker uses humor and sarcasm to underscore the ridiculousness of the situation while urging the company to reconsider its approach. The message is clear: Anthropic’s current policies are not only unfair but also damaging to its relationship with users, and the company needs to make substantial changes to regain credibility and support within the engineering community.