AI spending report reveals which tools businesses actually use, “vibe coding” finding success

Andreessen Horowitz (a16z), one of Silicon Valley’s most influential venture capital firms, recently published its first AI Application Spending Report in collaboration with Mercury, a fintech company serving startups. The report analyzed actual spending data from Mercury’s customers to identify the top 50 AI-native companies—startups built specifically around artificial intelligence rather than traditional companies that have added AI features.

This data provides a rare glimpse into which AI tools are actually driving business value rather than just generating headlines. While infrastructure providers show what capabilities companies are building, these application companies reveal where AI is being deployed in real products and workflows—a crucial distinction for understanding the technology’s practical impact.

The predictable leaders and surprising underdogs

OpenAI and Anthropic claimed the top two spots, which surprises no one given their prominence in generative AI. Other recognizable names populated the list, including ElevenLabs at #5 for AI voice generation, Perplexity at #12 for AI search, Canva at #17 for design automation, Grammarly at #22 for writing assistance, and Midjourney at #28 for image creation.

However, the most revealing insights come from the lesser-known companies making the list. These specialized tools illuminate how businesses are actually integrating AI into daily operations rather than pursuing flashy, headline-grabbing applications.

Customer service automation dominates this category, with three companies—Lorikeet (#8), Retell (#16), and Crisp (#46)—all providing AI-powered customer support platforms. Metaview (#19) offers AI assistance for recruitment processes, while Crosby (#27) operates as an AI-assisted law firm helping enterprise clients reduce time spent on routine legal tasks.

This pattern reflects what Forrester Research, a market analysis firm, recently characterized as AI’s transition into a “frumpy but functional” era. Rather than pursuing grandiose applications, businesses are increasingly deploying AI for mundane but valuable workflow improvements that deliver measurable returns.

The rise of vibe coding platforms

Third place went to Replit, an AI coding platform that represents one of the report’s most significant trends. Replit recently launched an agent designed to help non-programmers write and edit code using natural language prompts—an approach commonly called “vibe coding” because users describe what they want in conversational terms rather than technical specifications.

Three other vibe coding platforms made the list: Cursor (#6), Lovable (#18), and Emergent (#48). This strong representation signals growing enterprise adoption as these tools make custom AI application development accessible to business users without programming backgrounds.

The technology isn’t perfect—Replit made headlines when a coding error accidentally deleted a company’s entire database. However, spending patterns suggest businesses are finding these tools increasingly reliable for practical applications.

Replit’s higher ranking likely stems from its comprehensive approach. Unlike platforms focused solely on front-end design, Replit enables development of enterprise-grade applications, agents, and automations while providing the safety and governance controls that businesses require.

The vibe coding market could evolve in two directions: fragmentation into specialized platforms for different application types, or consolidation around a single dominant enterprise platform. Current spending patterns suggest the market remains open for either outcome.

Horizontal versus vertical AI applications

The report revealed a clear preference for horizontal AI applications—tools that can be deployed across entire organizations rather than serving specific job functions. Sixty percent of the top 50 companies offer horizontal services, while 40% provide vertical solutions targeting particular roles or departments.

Horizontal applications include foundational AI models like those from OpenAI and Anthropic, search enhancement tools like Perplexity, and productivity platforms like Merlin AI (#30). Meeting support tools showed particularly strong representation in this category, with Fyxer (#7), Cluely (#26), and Happyscribe (#36) all making the list.

This preference for horizontal tools aligns with recent MIT research suggesting that 95% of business AI implementations fail when companies take top-down approaches. The study found that successful AI adoption occurs when individual employees and teams can experiment with tools that address their specific needs rather than having AI strategies imposed from above.

Among vertical applications, the report’s authors identified two distinct categories: tools that assist human teams and those designed to replace them. Current spending patterns strongly favor assistance over replacement, suggesting businesses prioritize augmenting rather than eliminating human capabilities.

Implications for workforce strategy

These spending patterns offer insights into how AI might affect employment in the near term. Major employers are publicly emphasizing workforce augmentation over replacement. Doug McMillon, CEO of Walmart (the world’s largest private employer), recently stated that while AI will “change literally every job,” the company remains committed to avoiding layoffs and instead upskilling existing employees.

A survey by Creatio, a customer relationship management platform, found that 84% of executives expect AI tools—particularly AI agents that can handle complex tasks—will empower rather than replace current employees and potentially create new job categories.

However, these public statements should be interpreted carefully. Business leaders have obvious incentives to avoid announcing layoff plans, and attitudes toward automation may shift as the technology becomes more capable. The current preference for augmentation tools may reflect AI’s current limitations rather than permanent strategic choices.

What this means for business strategy

The a16z spending report suggests that successful AI adoption follows a pattern of targeted, practical applications rather than transformative overhauls. Companies are investing in tools that address specific workflow inefficiencies—customer service bottlenecks, recruitment processes, legal document review, and meeting management—rather than pursuing AI for its own sake.

This approach aligns with broader technology adoption patterns where incremental improvements often deliver more value than revolutionary changes. For business leaders evaluating AI investments, the report suggests focusing on horizontal tools that can scale across departments and vertical solutions that clearly augment rather than threaten existing team capabilities.

The prominence of vibe coding platforms indicates that AI’s democratization of technical capabilities may be one of its most significant business impacts. When non-technical employees can create functional applications through natural language instructions, organizations can respond more quickly to operational challenges without traditional development bottlenecks.

As AI tools mature and prove their value in these practical applications, businesses that have built familiarity with targeted implementations will likely be better positioned to adopt more sophisticated capabilities as they emerge. The companies succeeding in a16z’s spending rankings are those solving real problems rather than chasing technological novelty—a pattern that suggests the AI market is beginning to mature beyond its initial hype phase.