The most-used AI tools according to 2,500 companies

According to our anonymized Pilot dataset of 2,500 businesses, the most-used AI tools are OpenAI, Anthropic, Cursor, Midjourney, and Perplexity. 

Note: This list is based on the percentage of businesses that spend something on these tools, not how much they spend. It’s just a measure of popularity.  

Below, we analyze what we can see in this data. At the end, we share ideas for how this might influence the tools you choose. Or, if you run an AI startup, ways you might change your marketing.

1. Market share for the top three tools is over 100%

If 63% of businesses pay for OpenAI, 36% pay for Anthropic, and 34% pay for Cursor, that means a significant number of companies are using more than one tool. This tracks with what we observe talking to owners and founders: Many people are still busy experimenting, and sometimes find that a general LLM tool like OpenAI can’t do it all.

If you run an AI startup, you know this difficulty all too well: Customers are quick to adopt and try an AI tool, but equally quick to drop it. Churn rates are high, competition is fierce, and buyers are fickle. (By the way, you should use these three usage and financial metrics to make your case when raising a round.)

Curious which type of AI startups pay their founder the most? → 

2. Each category has an emerging leader

Each category such as coding or generating images seems to have 1-2 clear leaders, and there seems to be plenty of room use for use case-specific applications, like creating AI voices.

Of course, these lines are blurry. General LLMs such as ChatGPT and Claude can generate images just like Midjourney, listen in on calls like Fireflies, and more. 

3. The AI tool market is still growing rapidly

The big, newsworthy launches are mostly behind us for now. AI app users are now accustomed to tools, aren’t surprised at most AI capabilities, and have mostly formed their preferences.

In the early days, talking to business owners, people were fretting about whether to switch from Claude to OpenAI, or OpenAI to Gemini, on a weekly basis. But increasingly, those general LLMs all have all settled into offering similar features.

The last really big change was March of this year, when Anthropic released Claude 3 and saw a 574% month-over-month growth spike, jumping from 1.7% to 11.2% adoption. This included a little-known feature called Claude Model Context Protocol (MCP) that for those in the know, helps you connect anything to Claude and have its agents run your software for you. Partly as a result, it’s become a developer favorite, and one-third of businesses in our dataset use Claude.

4. Should business owners use a general LLM or a bunch of purpose-specific tools?

This is a complicated question and our data doesn’t really offer a clear answer. What does seem clear is that the AI tool barrier, for many business owners, is how much onboarding and work they require. As easy as it seems to copy-paste sales transcripts into Claude, those little tasks add up, and busy founders can’t afford to be doing all that clicking. This means it’s not a bad idea to look for AI tools that fit into your workflow and actually do work, not just respond to your queries. For example, choosing to standardize on Google’s Gemini because it lives within your Google Docs, Google Slides, and Gmail. 

When building your AI tech stack, the question is really, what saves you actual time?

For more data updates, visit our insight portal