After hearing good things about Anthropic's Claude AI, I wanted to give the chatbot a try for myself. I'm usually skeptical of LLM tools because of their habit of hallucinating, so I wanted to see if Claude really offered a different experience.

While I was aware that Claude had limits for users on its free and paid plans, I didn't realize just how restrictive they could be. This meant that I almost gave up on using the tool entirely.

My first session with Claude was extremely limited

I started using the tool as Anthropic ramped up peak-usage limits

Limits for free users of AI tools are nothing new, but I didn't realize how restrictive Claude would be in comparison to others. Within two prompts, I had reached my free message limit for five hours — meaning that I would need to upgrade to a paid plan or wait until five hours elapsed if I wanted to continue.

I was testing different AI tools to help me with my taxes, and purposefully stuck to free plans so that I could compare the models available to most casual users. However, Claude was the only tool in which I encountered usage limits. When my free message quota refreshed, it was used again within two prompts.

This meant that what took 30 minutes to an hour in other tools took multiple days in Claude. I realized that I wasn't the only one experiencing this issue, as users had taken to the Claude subreddit to complain about how quickly session limits were taking effect.

Thariq Shihipar, a member of the technical staff at Anthropic, posted on X that five-hour session limits had been adjusted during peak hours. This meant that free, Pro, and Max users would move through their session limits faster from 5am to 11am PT. According to Shihipar, the change was made to manage the growing demand for Claude.

I was not alone in my frustration. And it made me wonder whether it was worth bothering with the tool at all.

Claude's quality meant that I persisted

I also tweaked my usage behavior

I eventually completed my test of Claude during off-peak hours, where I didn't encounter the annoying limits anymore. I was able to finish the prompts I had prepared, as well as add some follow-up prompts.

Notably, in my test on my tax spreadsheet, Claude was the only tool that managed to get my income correct. This immediately made me pause and reconsider my plan to ditch the tool as soon as I was done with my test. As a result, I decided to test it for a few other things.

In addition to the quality of its responses, other factors won me over. When it comes to visualizing data and creating charts, spreadsheets, and checklists, Claude goes the extra mile. For example, its ability to color-code my spreadsheets is helpful when dealing with a lot of data. When it comes to its responsive visualizations, I can download them as HTML files so that I can view them in my browser whenever I need them.

I also enjoy its Projects feature, which is something that a tool like Gemini currently lacks. Projects are also useful for burning through usage limits more slowly, since Claude caches the data in your uploaded files.

When I used Claude to help me brainstorm more articles based on successful ones I'd done in the past, I found that its results were particularly impressive.

As a result, I'm sticking with the tool — though I'm saving my more intensive tasks for off-peak hours.

The backlash is understandable though

Especially for paying users

While I'm sticking with Claude, this doesn't mean I don't understand the frustration that users are feeling. I am a freelancer and I work from home, so I am able to simply shift my hours around to use Claude during off-peak periods. This isn't the case for many people.

I think I'd also be more frustrated if I was a paying user. I am entertaining the idea of a Claude Pro subscription, but I think peak time usage limits would sting a lot more if I was being told when I could effectively use a product I was paying for.

I also don't have a particularly intensive workflow — I mostly use AI tools for brainstorming. However, peak time usage limits can be detrimental to heavier workflows, especially with Claude's propensity to provide comprehensive answers. This makes it more difficult to refine prompts, since you have to preempt the chatbot's output and what it might overlook. At the same time, you have to make sure your prompts aren't unnecessarily long, since this also impacts usage limits.

Hopefully Anthropic will find a way to provide a smoother experience

I'm hoping that the current restrictive limits on Claude will smooth out to provide a better experience for users. I feel like I've started using the tool at the worst possible time, even though I am impressed by its results.

For longtime users and those who pay a subscription, I imagine the change is even more frustrating. Claude is extremely promising — it would be a pity if it loses footing in the market due to people being unable to actually use it.