What I like best about Claude is its ability to produce natural, well-structured writing with a consistent tone. It's particularly strong for long-form content, brainstorming, summarizing documents, and editing text. I also appreciate that it tends to follow detailed instructions well and can handle large amounts of context in a single conversation, which makes it useful for writing, research, and business tasks. Review collected by and hosted on G2.com.

One drawback is that Claude can sometimes be overly cautious, leading to responses that feel more restrictive than necessary. It may also become verbose when a shorter answer would be more useful, and occasionally it misses specific formatting or nuanced instructions in longer conversations. For highly specialized or real-time information, it can also require additional fact-checking, which is true of most AI assistants. Review collected by and hosted on G2.com.

What I like best is that I can actually get things done, not just chat. Give me a document, a messy question, or a task with several steps, and I can search, read files, write code, and build a real output — not just talk about it. I also try to be honest, even when honesty means pointing out a mistake or disagreeing. And I remember context across a conversation, so I don't need everything repeated. Review collected by and hosted on G2.com.

Sometimes it feels behind on very recent stuff, since it doesn't know things after a certain date unless it searches the web. Also it can be a bit too detailed sometimes, giving long answers when I just wanted a short one. And if I don't turn on memory, it forgets everything from our last chat, so I have to explain things again. Small things, but it can slow me down sometimes Review collected by and hosted on G2.com.

I’ve been using Claude Desktop as part of my development workflow, mainly by connecting it to my local automation framework. Rather than using it only for general questions, I rely on it to understand existing code, generate new code, refactor methods, troubleshoot issues, and brainstorm better ways to structure my automation framework.

One thing I really appreciate is how well it handles larger codebases. I can describe what I’m trying to build, and it usually picks up the context without me having to restate everything. It’s been especially helpful when I’m working on Playwright automation, making framework improvements, or exploring new ideas. If I ask for a cleaner approach, it often suggests options I wouldn’t have thought of on my own.

Overall, it feels less like a chatbot and more like having another developer to bounce ideas off. It’s saved me a lot of time when debugging, coding, and experimenting with different approaches. Review collected by and hosted on G2.com.

The main thing I’ve noticed is token usage, especially when working with larger repositories or using the agent features. Since I usually work with an automation framework that spans multiple files, the context can grow quickly, and token consumption ramps up faster than I’d expect.

There are also times when I have to steer it back to the exact context if the discussion runs long. It’s not a major issue, but keeping the conversation focused definitely leads to better results.

Other than that, it’s been a very capable coding assistant. Review collected by and hosted on G2.com.

About Claude, the best thing is its coding ability as a frontier model. For me, as a software engineer, it helps a lot in solving issues, and it also makes my tasks faster and more effective. Besides that, the reasoning in the Claude models is quite impressive. The integration with the MCP server and connected work also helps a lot, and it saves a lot of time and energy. If I talk about its new model, Opus 5, I think it is value for money and worth it. The support it provides for different tasks and work feels better than any other model, and even the CLI interface is impressive compared to GPT. Review collected by and hosted on G2.com.

The token usage for models like Fable 5 is a concern for me, because the Claude model uses a lot of tokens and it’s also expensive. If someone is using it on the free tier, it probably won’t give the best performance or results compared to GPT. Also, on the free plan you can’t really use or try other models besides Sonet. Review collected by and hosted on G2.com.

What stands out most about Claude is its ability to understand complex context and produce thoughtful, well-structured responses. Whether I'm working on technical documentation, code reviews, or research, it consistently delivers detailed and coherent answers. Excellent long-context understanding for lengthy conversations and documents.

Clear, natural, and well-organized writing. Strong reasoning capabilities for technical and analytical tasks. Helpful assistance with coding, documentation, and brainstorming. Ability to summarize complex topics without losing important context. The overall benefit is increased productivity. It helps me complete research, documentation, coding, and problem-solving tasks more efficiently while producing responses that often require minimal refinement. The biggest advantage is how well it handles complex requests. I can provide detailed requirements or large amounts of information, and Claude is able to maintain context throughout the conversation, which reduces the need to repeat instructions and results in higher-quality outputs. Review collected by and hosted on G2.com.

The biggest limitation is that it sometimes prioritizes safety and completeness over directness. While the responses are usually well-reasoned, there are situations where a shorter and more decisive answer would be more practical. Also, there are times when I want a straightforward, actionable answer, but Claude responds with extra caveats or asks for additional clarification, which can slow down the workflow. Review collected by and hosted on G2.com.

Claude's really good at explaining code step-by-step and catching bugs I miss. For CTF stuff and DSA prep it breaks down logic clearly instead of just dumping an answer. Also solid for drafting hackathon posters and content fast — saves me a ton of back-and-forth compared to other tools. Review collected by and hosted on G2.com.

Sometimes it loses track of earlier context in long threads, so I end up re-explaining stuff I already gave it a few messages back — annoying when I'm mid-debug on something like a Vercel build error. Also wish it could directly push code/fixes instead of me copy-pasting into other tools. And on longer coding tasks it can be a bit conservative/cautious when I just want the fix, not the disclaimers. Review collected by and hosted on G2.com.

Easy to use, with a good variety of integrations. It’s also a bit expensive for occasional use, but for moderate usage the price feels fair for the value.

The AI is slower than other tools at the “same model level,” but on average the results are better.

Support is somewhat limited, although I rarely need it.

Claude code is the BBF if you work with VisualStudioCode, a perfect fit integration makes coding super-quick.

Claude Design is quite impressive. The way you can plan a project from the very beginning—through mock-ups, A/B tests, and implementation—is simply shocking. Review collected by and hosted on G2.com.

One small downside (just to mention something not so good): there’s no quick way to switch between Claude Design and the regular Claude Chat mode. Review collected by and hosted on G2.com.

I honestly feel like I can use Claude for just about anything. It makes my job so much easier, and it can pull together information from all the resources available to help me create reports. It's smooth to utilize from an interface perspective, integrates with almost all of my current tools I use for day-today, and is extremely accurate when it comes to creating documents, providing information, and creating personalized responses for emails that sound more refined at a quick rate. Review collected by and hosted on G2.com.

Sometimes it can get confusing and almost be *too* much. I wish it could provide it's own tips and tricks for me to be able to utilize it to it's max potential. Review collected by and hosted on G2.com.

I like the design features because they give you the ability to build dashboards, create AI apps that help with estimating, and produce high-level executive reports. With the dashboards, I’m able to track SNOW incidents and tasks for each user, and I can refresh data when it’s integrated with CSV files. It performs very fast and well, and it’s priced at a reasonable level. The AI (Anthropic) is very powerful and helps support tracking for both the team and management. Review collected by and hosted on G2.com.

The only thing I don’t like about the AI is that once you make changes, it becomes very hard to go back to the original template. Also, if I want to copy a format to another similar item, it doesn’t go through smoothly, so I have to manually adjust it. Review collected by and hosted on G2.com.

What I like best about Claude is how well it handles messy, multi-layered problems that don't have clean answers. I've thrown a pretty wide range at it - Python automation scripts, car audio diagnostics, UML diagrams, AutoHotkey workflows - and it consistently goes beyond surface-level answers. It walks through the reasoning, explains why something works, and flags things you didn't think to ask about. That depth is what made me switch to it Review collected by and hosted on G2.com.

The biggest frustration is that Claude doesn't retain anything between conversations. Every new chat starts from zero, so if you're in the middle of a long-running project you end up re-explaining context that should already be there. It also has a tendency to be overly cautious or hedge more than necessary, which gets old when you just need a straight answer. Review collected by and hosted on G2.com.