In recent years, venture capitalists have been investing heavily in artificial intelligence companies, reflecting the technology's growing influence in Silicon Valley and across the globe. However, not every AI startup is capturing the interest of investors.
Although many companies are rebranding with "AI" in their titles, certain startup concepts are losing favor among investors. A discussion with venture capitalists revealed trends regarding which AI software-as-a-service (SaaS) ventures are currently undesirable.
Aaron Holiday, a managing partner at 645 Ventures, noted that investors are keen on startups focused on creating AI-native infrastructure, specialized vertical SaaS with unique data, systems that facilitate user tasks, and platforms integrated into essential workflows. Conversely, investors are becoming less interested in companies that merely add thin layers to existing workflows, offer generic horizontal tools, provide lightweight product management solutions, or deliver surface-level analytics—essentially, services that could be automated by AI agents.
Abdul Abdirahman from F Prime pointed out that vertical software lacking proprietary data advantages is no longer appealing. Igor Ryabenky, managing partner at AltaIR Capital, emphasized the need for significant product differentiation. He argued that if a company’s unique offerings are primarily in user interface improvements and automation, they may struggle to succeed in the current market.
"The barrier to entry has lowered, making it increasingly difficult to establish a strong market presence," he explained. New entrants should prioritize genuine ownership of workflows and a clear comprehension of problems from the outset. Instead of relying on extensive codebases, speed, focus, and flexibility in pricing strategies—away from rigid per-seat models—are essential for success today.
Jake Saper, general partner at Emergence Capital, pointed out trends in workflow ownership. He identified distinctions between Cursor and Claude Code, stating that ownership of the developer's workflow is becoming increasingly important, as many developers prefer execution capabilities over a detailed process.
Saper warned that products focusing on "workflow stickiness" face challenges as agents take over tasks historically handled by humans. He remarked that previously, getting users to engage with a product was a significant advantage; however, if AI agents can perform tasks, human-driven workflows lose their significance.
Furthermore, integration solutions are becoming less desirable as technologies like Anthropic's Model Context Protocol (MCP) streamline connections between AI models and external systems, reducing the necessity for downloading multiple integrations or creating custom solutions.
Transitioning into the future, Abdirahman noted that tools for workflow automation and task management will likely diminish as agents increasingly handle those tasks, referencing the performance of public SaaS companies that are losing footholds as AI-native competitors emerge with superior technologies.
Ryabenky highlighted that SaaS companies struggling to attract funding are often those whose offerings are easily replicated. Generic productivity software, basic customer relationship management systems, and shallow AI integrations face risks because strong AI-focused teams can quickly recreate them.
Ultimately, investors are inclined toward SaaS solutions that possess depth, expertise, and integration into crucial workflows. Focusing on deep AI integration into products and adjusting marketing strategies to highlight these advancements will be critical. Ryabenky concluded that capital is shifting towards businesses that have control over workflows, data, and specialized knowledge while moving away from easily copied offerings.