TechWithDavid
TechWithDavidApr 3
Tech

Claude Code is BROKEN (And Anthropic is Blaming YOU)

9 min video5 key momentsWatch original
⚡
TL;DR

Anthropic admits to tighter usage limits on Claude, attributing issues to user error rather than bugs.

Key Insights

1

User error — Anthropic's response shifts blame to users, suggesting they use the product incorrectly.

2

Claude's usage limits have been reduced by up to 75%, impacting many users.

3

Transparency needed — TechWithDavid calls for better communication and transparency from Anthropic.

4

Model switching — Practical tips include switching models and managing context to optimize usage.

5

Frustration with updates — Users are frustrated with the lack of clear updates on system performance.

Deep Dive

Anthropic's Acknowledgment

Anthropic has acknowledged the issues users are facing with Claude, citing tighter peak hour limits and some efficiency fixes. They claim these changes are not bugs but rather adjustments to manage increased demand.

User Frustrations

TechWithDavid expresses skepticism about Anthropic's claims, noting his own usage dropped significantly without any changes to his workflow. He argues that the company's communication is inadequate, leaving users in the dark.

Suggested Solutions

To mitigate issues, users are advised to switch to smaller models, lower effort settings, and manage their context windows effectively. TechWithDavid emphasizes that these adjustments might lead to lower quality but could help manage usage.

Call for Transparency

David stresses the need for Anthropic to provide clearer insights into usage limits and scaling issues. He believes that without transparency, user dissatisfaction will continue to grow.

Takeaways

  • ✓Monitor your usage and consider switching to smaller models.
  • ✓Reduce context window to save on tokens.
  • ✓Engage with community feedback for additional tips.
  • ✓Advocate for better communication from service providers.

Key moments

1:00Anthropic's Response

“We've investigated and we're sorry this has been a bad experience.”

2:00User Experience

“I personally saw my own usage drop by about three quarters during this period.”

4:00Need for Transparency

“I would have much preferred radical transparency if they had just come out and said we cannot keep up with demand.”

6:00Practical Tips

“You can switch using /model where they recommend Sonnet and you can also use Haiku.”

8:00Future Considerations

“If we're still in a ridiculous situation a few weeks from now, I'll probably try Gemini Pro or Codeex again.”

Get AI-powered video digests

Follow your favorite creators and get concise summaries delivered to your dashboard. Save hours every week.

Start for free