AI is changing the skills people have, what people do, and what their teams look like. The change is hitting large companies that are different from small and medium-sized businesses. Large companies are gaining progressive benefits from their existing team structures, but startups are emerging with true AI native companies built around the next generation of software.
But what will this next-generation software look like? We collaborated with Mercury, an ambitious company at Fintech, to collect and analyze spending across a database of over 200,000 customers, which we used for banks*, credit cards and all financial workflows, and examined patterns over the past three months (June to August 2025). Next, we identified the top 50 AI-Native Application Layer Companies – Similar to the top 100 GEN AI consumer apps, it is built around spending data and web traffic data.
Unlike infrastructure providers where startups are enabled (calculation, models, developer tools), these companies show where AI is actually applied in products and workflows, and their distinction is important. The findings draw similarities related to Mercury's broader survey of non-mercury customers, indicating that most early stage companies surveyed plan to increase AI spending, with people using AI seeing higher ROI than traditional tools.
Companies that created the list ranged from vibe coding platforms to creative tools and customer service solutions. And rankings bring real evidence to what AI enthusiasts are intuitive. Not only does AI augment employees in specific roles (“AI for patent lawyers”), it also transforms certain roles into skills that are widely deployed across the company (everyone becomes creative!).
Methodology Notes – It was not included primarily companies selling cloud services (Ex. Azure) or GPU (Ex. CoreWeave) or infrastructure tools. Google includes both Google Cloud and Gemini models. Because I couldn't solve these two.
This data is limited to transactions made through mercury such as ACH, IO card spending, wires and more. It does not capture Mercury customer spending to other sources, including non-mercury cards that may be refunded. The data does not reflect expenditures from private mercury customers.
1. The cost of horizontal applications is slightly numerous.
Divide applications into horizontal applications (focused on increasing overall productivity across the enterprise and available to everyone) and vertical applications (targeting a specific role). Horizontal companies account for 60% of the list, but 40% of vertical companies.
Within horizontal applications, the most well-known name is the common LLM assistant. #1 Openai, #2 Humanity, and #12 Confusing, #30 Merlin AI follows. Additionally, there are workspaces that allow users to access the LLMS context of existing files. The fact that several products in this category created ranks indicates that the race here will not win. Haven't seen the appearance of a true winner yet, or this space will not “winners take everything away.” Users switch between different UIs (or models) depending on their needs.
Beyond the general assistant, another major category of horizontal products met support. These are primarily attracting attention – #7 Fyxer (also draft email), #36 Happyscribe, #38 Plaude (manufacturer of hardware pins), #41 Otter AI, and #49 Read AI. However, this category is also expanding with products such as #26 Cluely, which offers real-time in-meeting feedback.
Are the other two application categories we argue currently level? Coding creative tools and atmosphere! These were previously domains of specific functions. The marketing or design department used creative tools, while the engineering team was drawn to atmosphere coding. However, AI has launched applications in these categories that can (and can be used) by people of all roles. This is usually seen in several categories where domain-specific tools are more horizontal.
Creative tools are represented in the list with 10 names and ranked as the largest single category. The leader is the all-in-one suite Freepik (#4), followed by speech generator ElevenLabs (#5), from the text. Images and videos had the most applications in rank, including Canva, Photoroom, Midjourney, and Description, Opus Clip, Capcut, and more, respectively. However, avatars have also emerged in #47 Arcads (used primarily for advertising) and #50 Tavus (multipurpose).
2. Vertical applications can be an augerment or alternative.
There are two ways to think about vertical AI. On the one hand, reducing the most repetitive and mediocre tasks allows human employees to be recharged into roles and spend time on valuable work. Meanwhile, the need for human teams can be reduced or eliminated.
So far, we are primarily looking at the former. Of the 17 vertical application companies, 12 focus on human overload, while five aim to act as “AI employees” completing end-to-end workflows. The latter include #27 Crosby Legal (Agent Law Firm), #34 Cognition (AI Engineer), #37 11x (Automatic GTM Employee), #39 Serval (AI IT Service Desk), and #42 Alma (AI-Powered Mivirus Law Services).
Nevertheless, we look forward to the emergence of more end-to-end agent products, as well as AI-Native Services Business. These could be exemplified in Mercury's client base, particularly as new startups may not be trapped in multi-year contracts with expensive providers such as lawyers, accountants and more.
What about the most popular categories of vertical software? Several representatives were determined for each of the three verticals.
- Customer Service: lorikeet (#8), customer.io (#14), ada (#40), crisp (#46)
- Sales / GTM: installly(#13), clay(#25), 11x(#37)
- Recruitment / HR: Micro1 (#9), MetaView (#19), praise (#43)
Furthermore, the two companies belong to the wider “operation” bucket. These include Delve (#11, Compliance Automation) and Combinely (#29, Accounting).
3. Atmosphere coding is more than just a consumer trend. I landed at work.
One category that has made a particularly strong enterprise migration is vibe coding. Four companies targeting AI app building, whether engineers or non-engineers, have created the list. Also, beyond the two biggest model companies (Openai and Humanity), the top product on the list was Replette (#3). Agent product development tool.
To us Consumers Top 100 lists based on web traffic (not revenue), Lovable went into the top quarter, while Replit was further down the list. This order was reversed on Enterprise List. Lovable created a list, but Replit generated around 15 times more revenue than Lovable from Mercury customers we surveyed.
Why does that happen? There are several theories. While Lovable focuses more on rapid UI and component generation (providing a low barrier to pure consumer entries), replicas go beyond front-end design by enabling the development of enterprises, fully functional apps, agents, and automation. The agent can run autonomously for hours. It also provides embedded cloud services such as databases, authentication, and secure publishing directly within the platform. When combined with Replit's enterprise control, it means that feature sets often make more sense within the enterprise.
We are interested in observing the evolution of atmospheric coding over time. With the rise of platforms for developing different types of applications, will space “fragment”? Or, at least in the enterprise realm, does one winner “get the most” and the space starts to converge?
4. Products are moving from consumers -> Prosumer-> Enterprise.
The majority of companies on the list (nearly 70%) can be hired by individuals and brought to the team. You do not need to use an enterprise license. In fact, 12 companies ranked Also It has appeared in the latest Consumer Top 100, which mainly tracks traffic from B2C AI products.
Of these, 11 companies It's begun It has evolved over time to provide team or enterprise functionality, primarily as individual products. Moreover, some of them still generate revenue for the majority of consumers (Cluely, #26, and Midjourney (#28) are two such examples). This also applies to many of the biggest model companies, such as Openai. As of October last year, 75% of revenue came from consumers, and is estimated to be close to the recent 50/50 split.
What's going on here? I wrote about this trend in “The Great Exluince: A New Era of Consumer Software.” Thanks to AI, consumer products are more powerful than ever, but at the same time can serve enterprise use cases. Given the urgency of hiring AI to improve employee efficiency, these tools are “pull into the enterprise faster than ever.
In contrast to the previous era of software, it excels in product-driven growth and driving the advanced market has become a viable feat in the first year or two of the company's life. We hope that more companies will start consumer-first in this future edition of the enterprise list.
*Mercury is a financial technology company, not a bank. Business banking services provided through Choice Financial Group, Column NA, and Evolve Bank & Trust. Member fdic. Personal banking services provided through Choice Financial Group. Member FDIC. IO cards are issued by Member FDIC of Patriot Bank, subject to MasterCard® license.
