a16z AI Application Spending Report: What Enterprise AI Apps Are Companies Actually Paying For｜Hello Tech
Where the Money Is Going: The Future Direction of AI Applications
Generative AI is reshaping the skills people need, the nature of work itself, and how companies structure their teams. Inside large corporations, AI is being deployed as an incremental upgrade on top of existing organizational frameworks. But in the startup world, a new breed of truly AI-native companies is emerging — built from day one on next-generation software.
So what exactly is this "new generation of software"? In a report produced with fintech company Mercury, a16z analyzed AI spending patterns across more than 200,000 companies in Mercury's customer database during June through August of this year, filtering for those with the highest and most representative AI expenditure to produce the Top 50 AI-Native Application Layer Companies.
Unlike infrastructure providers, these companies demonstrate AI's real-world implementation at the product and process level. This is a live signal of where AI spending is actually flowing*.
The list spans everything from vibe coding to creative tools to customer service solutions. The ranking also validates a trend much discussed in AI circles: AI won't merely assist employees in specific roles — it will become productivity infrastructure deployed company-wide.
Follow the money, and you follow the future. In a sense, this report is a "map" of where AI applications are heading.
Compared against a16z's earlier Gen AI Top 100 Consumer Apps, some intriguing patterns emerge. In the Gen AI era, the path from consumer to enterprise has become remarkably smooth. Some AI companies have managed to traverse the journey from product-led growth (PLG) to enterprise markets in remarkably short order. Going forward, we may see growing numbers of companies that start consumer-facing and later expand into B2B.
Here are four key trends worth noting from the rankings.
01
Horizontal Applications Slightly Outspend Vertical Ones: Making "AI for Everyone" a Reality
a16z categorizes AI applications into two types:
Horizontal applications: Boost overall company productivity, usable by nearly anyone
Vertical applications: Tools designed for specific roles or functions
In this ranking, horizontal applications account for roughly 60%, vertical applications about 40%.
Among horizontal applications, the most familiar are general-purpose LLM assistants: OpenAI at #1, Anthropic at #2, Perplexity at #12, and Merlin AI at #30.
Beyond these, an expanding category of workplace platforms lets users invoke LLMs directly within their existing documents and tools, such as Notion at #10 and Manus at #33.
The sheer number of products in this space suggests competition in horizontal AI applications remains fierce. A definitive "winner" has yet to emerge — or perhaps this category won't see winner-take-all dynamics at all, with users fluidly switching between different interfaces and models based on their needs.
Beyond general assistants, another major horizontal category is meeting support tools. These are primarily meeting transcription apps like Fyxer at #7 (which also drafts emails), Happyscribe at #36, Plaude at #38, Otter AI at #41, and Read AI at #49. The category continues to expand its boundaries — Cluely at #26, for instance, provides real-time coaching during meetings.
What's notable is that tools once considered vertical solutions for specific functions, such as creative tools and vibe coding platforms, are also shifting toward horizontal applications. AI has broken down barriers. Now employees in every role can use these tools.
In creative tools, ten companies made the list — the most represented category in the ranking. All-in-one suite Freepik placed #4, text-to-speech generator ElevenLabs #5. Breaking this down:
-
Image and video tools dominated in volume, spanning image generators like Canva, Photoroom, and Midjourney, and video tools like Descript, Opus Clip, and CapCut;
-
Avatar tools are also gaining traction, such as Arcads at #47 (primarily for advertising scenarios) and multi-purpose tool Tavus at #50.
02
Vertical Applications: Augmentation vs. Replacement
Vertical applications follow two paths:
- Augmentation: Freeing employees in specific roles from repetitive tasks so they can focus on higher-value work
- Replacement: Directly assuming functions and reducing reliance on human labor
Currently, augmentation remains dominant. Of 17 vertical application companies on the list, 12 focus on improving employee efficiency, while 5 aim to become "AI employees" that autonomously complete full workflows.
These five "replacement-type" companies are:
-
#27 Crosby Legal (AI law firm);
-
#34 Cognition (AI engineer);
-
#37 11x (automated GTM employee);
-
#39 Serval (AI IT helpdesk);
-
#42 Alma (AI immigration legal services).
a16z expects more end-to-end agentic products to emerge going forward, potentially even AI-native service companies. Many newly established firms haven't locked themselves into multi-year contracts with lawyers, accountants, or consultants — they're more willing to "hire AI."
-
Customer service: Lorikeet (#8), Customer.io (#14), Ada (#40), Crisp (#46);
-
Sales/GTM: Instantly (#13), Clay (#25), 11x (#37);
-
Recruiting/HR: Micro1 (#9), Metaview (#19), Applaud (#43).
Additionally, several operations-focused tools: Delve (#11) in compliance automation, and Combinely (#29) in accounting automation.
03
Vibe Coding: From Consumer Trend to Workplace Reality, Enterprise Adoption Accelerating
Vibe coding has rapidly completed its leap from B2C to B2B. Four companies made the list: Replit, Cursor, Lovable, and Emergent. Replit stands out particularly — this agent-driven product development tool ranked #3 on the list, trailing only OpenAI and Anthropic.
In a16z's traffic-based Gen AI Top 100 Consumer Apps ranking, Lovable placed in the top quartile while Replit sat further back. On this revenue-based ranking, the situation reverses: Replit's revenue is roughly 15x Lovable's.
The likely explanation: Lovable focuses more on rapidly generating UI and components, with a lower barrier to entry suited for ordinary consumers. Replit, meanwhile, doesn't just let users build front-end UIs — it supports the construction of enterprise-grade, fully functional applications, agents, and automation workflows that can run autonomously for hours, with built-in databases, authentication, secure deployment, and other cloud services, plus enterprise access management. It's simply better suited for B2B scenarios.
Whether vibe coding fragments into platforms serving different niches, or (at least in B2B) converges around a few winners, remains to be seen.
04
Faster Trajectory: From Consumer → Pro Users → Enterprise
One striking finding: nearly 70% of companies on the list had products that started with individual users who then brought them into team and enterprise use — without requiring corporate authorization.
Twelve of these companies had previously appeared on a16z's traffic-based AI Top 100 Consumer Apps ranking. Eleven expanded from individual-focused beginnings into B2B offerings. Some still derive primarily consumer revenue, such as Cluely at #26 and Midjourney at #28. This pattern is pronounced even among large model companies: OpenAI, for instance, derived roughly 75% of revenue from consumers before October last year, a figure that has recently trended toward 50%.
Why? As a16z discussed in another article, The Great Expansion: A New Era of Consumer Software: AI is making consumer products powerful enough to simultaneously satisfy enterprise demands. With companies more urgently seeking to boost employee efficiency, such tools are being "pulled" into enterprise contexts faster than ever before.
In past software eras, the journey from PLG to enterprise might take years. In the AI era, some companies have navigated this path within one to two years of founding. a16z expects future enterprise rankings to feature growing numbers of companies that started consumer-facing and gradually expanded into B2B.
These spending patterns and trends offer valuable directional signals. Understanding where the money flows means better grasping the market, choosing directions, and allocating resources.
*This analysis excludes companies primarily selling cloud services (such as Azure) or GPU compute.