Investors have been dedicating substantial funds to AI companies in recent years as the technology gains traction in the tech industry and globally. However, not all AI companies are attracting investor interest.
Although many companies are rebranding to incorporate “AI” into their names, some startup concepts have fallen out of favor with investors. TechCrunch spoke with venture capitalists to understand which aspects of AI software-as-a-service startups are less appealing to investors now.
Aaron Holiday of 645 Ventures notes that popular SaaS categories include startups developing AI-specific infrastructure, vertical SaaS with unique data, systems of action aiding task completion, and platforms significantly integrated into essential workflows.
Conversely, he pointed out that startups focusing on thin workflow layers, general tools, light product management, and basic analytics are losing interest. Abdul Abdirahman from F Prime stated that generic vertical software lacking exclusive data moats is less attractive. Igor Ryabenky of AltaIR Capital also emphasized that investors are not keen on products lacking depth.
Ryabenky mentioned that mere differentiation in user interface and automation is insufficient, as entry barriers have decreased, complicating the creation of a strong moat. New entrants must now focus on workflow ownership and a profound understanding of the problem from the onset. Large codebases are less advantageous, with agility, focus, and adaptability taking precedence. Pricing models should be flexible, and consumption-based models are more favorable than rigid per-seat arrangements.
Jake Saper from Emergence Capital discussed ownership, highlighting differences between Cursor and Claude Code as indicative of a trend. One commands developer workflows, while the other merely executes tasks. Products reliant on “workflow stickiness” may struggle as agents assume workflows.
Saper added that integration tools are declining in popularity, especially as Anthropic’s model context protocol (MCP) streamlines connecting AI models to external data and systems, circumventing the need for multiple or custom integrations.
Furthermore, workflow automation and task management tools for human coordination become less critical as agents execute tasks. Ryabenky pointed out that SaaS companies facing fundraising challenges are often those that are easily replicable, lacking deep integration, proprietary data, or embedded process knowledge.
Overall, what is compelling about SaaS is depth and expertise, especially tools integrated into crucial workflows. Ryabenky advised companies to integrate AI deeply into their products and update their marketing accordingly. Investors are shifting capital towards businesses with ownership over workflows, data, and domain expertise, moving away from easily replicable products.