a16z Report Reveals the AI Tools Startups Actually Pay For

On Thursday, Andreessen Horowitz released its inaugural AI Spending Report in collaboration with fintech company Mercury. The report examines transaction data from Mercury to analyze the leading 50 AI-native application layer companies that startups are investing in, following a similar approach to the earlier Top 100 Generative AI Consumer Apps.

a16z partners Olivia Moore and Seema Amble note that the data reveals companies continue to integrate a variety of AI products for specific tasks, with emerging applications rapidly gaining and losing traction.

“There’s an explosion of tools available,” Amble observed. “The market hasn’t settled on just one or two options per category.”

The report also highlights substantial spending on "human augmentors" or "copilots" designed to enhance workforce productivity, indicating that startups are not yet prepared to fully transition to autonomous agent-based workflows.

“As computer interactions evolve and comprehensive end-to-end agent workflows become feasible, we anticipate a shift from copilots to more autonomous agent tools,” Amble explained. “This is especially likely as people show growing enthusiasm for testing these solutions.”

Unsurprisingly, major AI labs dominated the top positions, with OpenAI claiming first place and Anthropic following at number two. Vibe-coding tools also featured prominently, with Replit at number three and Lovable at number 18. Cursor secured the sixth spot, while Emergent ranked 48th. Cognition, which focuses on enterprise-oriented coding solutions like Devin and Windsurf, placed 34th.

In a16z's previous consumer-focused analysis, Lovable ranked higher than Replit based solely on user traffic, as many individuals used it for personal projects. However, startups currently allocate less spending to Lovable compared to Replit, partly due to Lovable's limited enterprise capabilities. The diversity of companies on the list suggests there is ample space for multiple players to thrive simultaneously.

Join 10,000+ tech and VC leaders for growth and networking at Disrupt 2025

Netflix, Box, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, Vinod Khosla — just a few of the 250+ industry leaders headlining 200+ sessions crafted to provide actionable insights for startup growth and competitive advantage. Don't miss TechCrunch's 20th anniversary celebration and this opportunity to learn from top tech innovators. Secure your ticket before the event opens to save up to $444.

Join 10,000+ tech and VC leaders for growth and networking at Disrupt 2025

Netflix, Box, a16z, ElevenLabs, Wayve, Hugging Face, Elad Gil, Vinod Khosla — just a few of the 250+ industry leaders headlining 200+ sessions crafted to provide actionable insights for startup growth and competitive advantage. Don't miss this chance to learn from leading tech voices. Book your ticket before doors open to save up to $444.

“The future of vibe coding remains an open question,” Moore commented. “Will the market consolidate around one superior platform, or will we see four or five major vibe coding companies serving different application needs? We don’t have a definitive answer yet.”

Moore also expressed surprise at startups adopting consumer-focused tools like CapCut and Midjourney for business purposes.

“We’re noticing that many consumer-focused companies are being rapidly pulled into the enterprise space,” Moore noted. “Their tools are so user-friendly that individuals adopt them personally and then introduce them to their teams and workplaces.”

Horizontal applications accounted for at least 60% of the companies listed, while vertical applications made up the remaining 40%. The most prominent vertical software companies specialized in three areas: sales, recruiting, and customer service. The report also found AI making significant inroads into sectors where previous startup generations had faced challenges.

“What used to be service firms or consultancies are now becoming software companies in the AI era,” Moore stated.

Amble cited Crosby Legal as an example—a tool that rapidly reviews legal contracts, replacing what would traditionally require consultation with in-house counsel and extensive research. She explained that most current tools function as co-pilots, helping employees make decisions more efficiently rather than replacing entire job functions with fully automated agents.

“As technology advances and we develop fully capable agent coworkers, we expect to see a transition from co-pilots to comprehensive end-to-end agents,” she said, adding that AI tools can perform tasks like outreach much faster than humans.

The list also included numerous note-taking applications such as Otter.ai, Read AI, and HappyScribe—with no single option dominating the category. This illustrates Amble’s point that no one product has yet captured the market; instead, startups are experimenting with different options to find what works best for them. This variety benefits employees, who can select applications that suit their working style rather than being limited to a standardized corporate solution.

Another key finding was the growing convergence between consumer and enterprise applications. People are bringing personal tools they use at home into their workplaces, while founders are leveraging their favorite consumer apps to build their businesses. Previously, there was a clear distinction between personal and professional software stacks for startups.

Amble and Moore pointed to Canva as an example—a popular consumer application that also serves a substantial enterprise user base. Canva took years to introduce an enterprise plan, but as individual and business use cases increasingly overlap, companies are more open to blending the two.

“Your total addressable market is no longer limited to one segment—you can target both consumer and enterprise users,” Amble continued. She suggested that companies developing these products may also accelerate their professionalization, establishing enterprise-focused teams in areas like sales, marketing, and support to generate business revenue faster instead of relying solely on individual consumers.

Moore and Amble expect the list to evolve significantly in the coming years. Established companies are integrating AI features to remain competitive, while new entrants continue to introduce innovative approaches.

“In this context, ‘legacy’ can refer to what was relevant just 12 months ago,” Amble remarked. “If we revisit this analysis in a year, will the same note-taking applications even appear? Or will we see an entirely new set of players?”

This article has been updated to accurately reflect the note-taking applications mentioned.