In partnership with
Hey {{first_name | there}},
So lately, there’s been an uproar around Claude’s usage limits.
Users are complaining that they’re getting “robbed”. Hitting caps in a few prompts. Watching their weekly quota disappear. Getting locked out while still paying.
And honestly, on the surface, it does feel like a bait-and-switch.
But here’s the thing: You’re most likely getting rate-limited not because of the plan but because of how you’re using it.
Yes, the limits feel inconsistent and poorly communicated. But at the same time, most users are still treating AI like an unlimited chat product when it actually behaves more like metered infrastructure.
And that gap between expectation and reality is exactly where all this frustration is coming from.
What you should do instead
Start a new chat for every new task. Long chats = higher token usage every time
Don’t use the most powerful model for everything. Use Sonnet for writing, edits, and simple tasks. Use Opus only for complex reasoning or debugging
Avoid peak hours when possible. Usage is tighter during 8 AM – 2 PM ET (weekdays). Try doing heavier work early morning or later in the evening.
Keep your context lean. Remove unnecessary files, instructions, and tools.
Be intentional with prompts. Clear, scoped prompts reduce back-and-forth.
Quick question:
Which topic would you like to learn more about?
🛠️ AI Tools Worth Checking Out
Other AI Updates You Should Know
Google just introduced a new memory compression algorithm called TurboQuant.
It reduces the amount of working memory AI needs while keeping performance intact. In simple words, models can “remember” more while using less compute.
If this actually makes it into production systems, it directly affects pricing and limits. The reason tools like Claude feel constrained is because running them is expensive. Reduce that cost, and suddenly higher limits or cheaper plans become possible.
Google also rolled out Gemini 3.1 Flash Live, a model focused on real-time voice interaction.
The key upgrade isn’t just better answers. It’s how natural the interaction feels. Faster responses, better understanding of tone, and the ability to keep up with longer conversations without breaking flow.
Anthropic is reportedly testing a powerful new AI model called “Mythos”, described as their most capable yet. Because it's so powerful, especially in cyber capabilities, Anthropic is planning to roll it out slowly and carefully, rather than releasing it all at once.
Why does this matter? Two things. First, a more powerful Claude model is likely on the horizon, which could mean significantly better performance on complex tasks.
Second, it signals that as AI gets more capable, the safety and security tradeoffs become harder to manage. Future models might come with more restrictions or staged access rather than being available to everyone immediately. The cybersecurity world is already rattled by the news, which tells you this model is genuinely in a different league of capability.
Funny timing on these two stories sitting right next to each other. One report says Anthropic's new Mythos model is so powerful it's spooking the entire cybersecurity industry. The next one shows Anthropic partnering with Accenture to build cybersecurity tools using that same Claude AI.
So the company whose model is seen as a threat is also selling you the armor against it.
The actual tool, Cyber AI, puts Claude to work across the whole security pipeline, spotting threats, responding faster, and keeping AI agents in check through something called Agent Shield.
The bigger point here is that AI is no longer just a productivity tool sitting in your browser tab. It's becoming the backbone of how companies protect themselves, which means how capable and trustworthy these models are starts mattering way beyond just getting better answers to your questions.
Wikipedia has made it clear that AI-generated content shouldn’t be used to write articles.
The reason is simple. AI can produce text that sounds correct but includes subtle inaccuracies, unsupported claims, or made-up details. For a platform that depends on verifiable information, that’s a dealbreaker.
This highlights a bigger issue. AI is improving rapidly in capability, but reliability and trust are still lagging behind.
Which is why we’re starting to see pushback. Not against AI itself, but against using it without verification.
Final Thoughts
All of this ties together more than it seems.
People are frustrated with limits because AI is no longer “infinite.” Companies are working on making it cheaper and more efficient.
At the same time, models are becoming powerful enough to create real risks. So new systems are being built to control and monitor them. And on top of that, the interface itself is changing from text to voice.
AI isn’t just getting better. It’s becoming infrastructure.
And once that happens, cost, control, and trust start to matter just as much as capability.
Let me know what you think.
Are Claude’s limits actually unfair, or are we just seeing the real economics of AI for the first time?
- Aashish
Attio is the AI CRM for modern teams.
Connect your email and calendar, and Attio instantly builds your CRM. Every contact, every company, every conversation, all organized in one place.
Then Ask Attio anything:
Prep for meetings in seconds with full context from across your business
Know what’s happening across your entire pipeline instantly
Spot deals going sideways before they do
No more digging and no more data entry. Just answers.