An honest Anysphere review (2025): The truth about the $9b AI coding tool

You’ve probably heard about Anysphere's AI coding tool, Cursor. It shot up to a wild $9 billion valuation with over a million users, quickly becoming a poster child for the AI boom. But if you look past the headlines, you’ll find a more complicated story filled with user backlash, some serious competition, and a few major risks. This review is here to separate the hype from what's really going on.

What is Anysphere (and its product, Cursor)?

Anysphere is the company that built Cursor, a code editor designed with AI at its very center. Think of it this way: instead of just being a plugin that adds AI features to your current editor, Cursor is a full-fledged fork of Microsoft's popular Visual Studio Code (VS Code). The whole thing was rebuilt to have AI woven into every part of the development process.

The company was started by a group of MIT classmates. They were first playing around with AI for mechanical engineering software, but after getting early access to OpenAI's GPT-4, they saw a bigger opportunity. They pivoted to software development and launched Cursor in March 2023. Their big idea isn't just about finishing your lines of code for you; it's about changing the way developers write, fix, and manage code altogether.

Cursor's key features

To get why there’s so much buzz, you have to look at what Cursor brings to the table. It wraps a familiar coding environment around some genuinely powerful AI features.

An AI-native editor on a familiar foundation

Probably the smartest thing Anysphere did was build Cursor on top of VS Code. This means developers get an interface they already know and can use all their existing extensions, which makes switching over pretty painless. But because they forked the code instead of just making a plugin, Anysphere has complete control. This lets them embed AI features much more deeply than an extension ever could, making for a smoother workflow with native Git and terminal access built right in.

A screenshot of the Cursor AI editor, which many Cursor reviews praise for its familiar VS Code-based layout, is shown in this Anysphere review.

Agent mode and autonomous coding

Cursor's most ambitious feature is called Agent Mode. The goal is for it to act like an autonomous junior developer on your team. You can give it a high-level instruction in plain English, like “add a dark mode toggle to the settings page,” and the AI agent gets to work.

1. It starts by analyzing your request and looking at your entire codebase for context.

2. Then, it maps out the code changes needed across different files.

3. It writes and edits the code itself.

4. Finally, it tries to verify the changes and will flag any errors it finds.

This whole process is powered by a Retrieval-Augmented Generation (RAG) system, which helps the AI actually understand your project's structure. This makes its suggestions a lot more relevant and less likely to be wildly off base.

Codebase chat and model routing

Cursor also has a chat panel where you can ask questions about your code. You can use "@" symbols to reference specific files or folders, giving the AI the exact context it needs. It can even search external documentation with "@Web" commands. On the backend, Cursor uses a smart model router that picks the best AI model for your request from over 25 different options, including OpenAI’s GPT-4o, Anthropic’s Claude 3 Opus, and Google's Gemini.

A screenshot of the Cursor UI highlighting the different AI models a user can select, a key factor in this Anysphere review of Cursor's pricing.

Business model and pricing

Anysphere grew incredibly fast using a freemium model, but its pricing has also been a major source of drama.

Subscription plans

Cursor has three main pricing tiers, and the cost mostly revolves around how many "premium AI requests" you get.

The model is built around a "fast" vs. "slow" request system. Pro and Business users get a certain number of fast, high-priority responses each month. After you use those up, your requests go into a slower, shared queue. This usage-based system can make costs hard to predict, which is a headache for active developers.

A screenshot of the Cursor website showing the different tiers and monthly costs for the new Cursor pricing structure, a key part of this Anysphere review.

The controversy around their pricing

While Anysphere's growth is impressive on paper, it hasn't been a smooth ride. A lot of users have voiced their frustration. Reviews on Gartner and posts on Medium talk about "unfriendly pricing changes" that annoyed many of its early, loyal users. This kind of feedback makes you wonder about their long-term customer retention. If users feel the pricing is unfair or always changing, they'll eventually look for alternatives, making that explosive growth tough to maintain.

Hype vs. reality

So, is Cursor the future of software development or just another overhyped tool in the AI bubble? The truth is probably somewhere in the middle.

The pros: It's fast, innovative, and a peek into the future

You can't deny that for a lot of tasks, Cursor is incredibly useful. It can definitely speed up development by handling repetitive boilerplate code and gives a pretty exciting glimpse of where AI-assisted programming is headed. Many users say the experience is intuitive and slick, making complex AI interactions feel like a natural part of the coding process.

The cons: Platform risk, shaky retention, and quality control

For all its strengths, Cursor is facing some big hurdles.

• Platform Risk: As many have noted in developer communities, Cursor is basically a well-designed layer on top of VS Code and third-party LLMs. This creates a huge risk. What happens when Microsoft, which owns VS Code and is OpenAI's biggest backer, decides to build similar AI agent features directly into its editor for free?

Cursor is basically a well-designed layer on top of VS Code and third-party LLMs. This creates a huge risk. What happens when Microsoft, which owns VS Code and is OpenAI's biggest backer, decides to build similar AI agent features directly into its editor for free?

Reddit

• Unproven Retention: The backlash over pricing shows that their user base might be fragile. Early adopters are great for initial buzz, but keeping customers for the long haul requires stability, trust, and predictable value, all areas where Anysphere has had some missteps.

• Quality Issues: For a tool with a $9 billion valuation, users report a surprising number of bugs. You can find forum discussions where people mention that core features like code search can be unreliable, and the overall polish has been a letdown for some in the community.

A cautionary tale: The risk of AI making things up

One of the most telling stories involved Cursor's own AI support agent. When users ran into a bug that kept logging them out, the support bot confidently told them it was because of a new "single-device policy." The problem? Anysphere later confirmed that policy didn't exist; it was a complete fabrication, an AI "hallucination." This incident is a perfect example of the reputational and real-world risks of deploying AI without having tight control over what it knows and says.

This video provides an inside look at how Anysphere's Cursor is rapidly growing as an AI code editor.

Key takeaways for picking any AI tool

The Anysphere story offers some important lessons for anyone thinking about bringing AI tools into their workflow, whether it’s for coding, marketing, or customer support.

1. Avoid getting locked into one platform. Tools that are just wrappers around other platforms, like Cursor is for VS Code, can put you in a tough spot. It's usually safer to pick tools that integrate with the systems you already use instead of forcing you to move to a whole new environment.

2. Demand transparency and control. The story of the hallucinating support bot shows what happens when an AI is a "black box." Your team needs to be able to control exactly what knowledge your AI uses and what it's allowed to do. Being able to test and simulate its behavior before it ever talks to a customer should be non-negotiable.

3. Look for predictable pricing. Billing models based on per-use or per-resolution fees make budgeting a nightmare. They basically penalize you for being successful. A clear, fixed pricing structure lets you scale without worrying about surprise bills.

These issues aren't just for AI coding tools. They’re important for any team looking to use AI, especially in high-stakes roles like customer support.

A better way to automate support with AI

For teams looking to sidestep these problems, eesel AI offers a more controlled and predictable way to bring AI into your customer service and internal support. It's a platform built from the ground up for control, integration, and stability.

• Seamless Integration: eesel AI doesn’t ask you to ditch your helpdesk. It works with it, offering one-click integrations for platforms like Zendesk, Intercom, Freshdesk, and Slack.

• Total Control: You get to decide exactly what information your AI uses, whether it's past tickets, your Confluence docs, or other knowledge sources. Our simulation mode even lets you test your AI on thousands of your past tickets, so you can see exactly how it will behave before it goes live.

The eesel AI simulation dashboard shows how the AI uses past tickets to predict its automation performance, a key feature in this Anysphere review's comparison.

• Predictable Pricing: eesel AI has straightforward plans with no per-resolution fees. Your bill won't jump up after a busy month, so you can grow your support operations without any financial surprises.

A screenshot of the eesel AI pricing page, which shows clear, public-facing costs, is highlighted in this Anysphere review.

Learn from Anysphere and choose your AI partner wisely

Anysphere's journey is a fascinating, if cautionary, story from the current AI gold rush. It shows off the incredible things AI can do, but it also serves as a warning about the risks of a "move fast and break things" mindset. When you're ready to bring AI to your support team the right way, with complete control, easy integration, and predictable costs, give eesel AI a try.