Investors have poured billions of dollars into AI companies in recent years as the technology continues to have an impact on the Valley and, by extension, the world. But not all AI companies are attracting investor attention.
In fact, while it seems like every company is rebranding these days with “AI” in their name, some startup ideas are no longer popular with investors. TechCrunch spoke to VCs to find out what investors are no longer looking for in AI software-as-a-service startups.
According to Aaron Holiday, managing partner at 645 Ventures, SaaS categories that are popular with investors today include startups building AI-native infrastructure, vertical SaaS with proprietary data, action systems (those that help users complete tasks), and platforms deeply embedded in mission-critical workflows.
But he also gave a list of companies that are considered very boring to investors these days. It's a startup building thin workflow layers, general-purpose horizontal tools, simple product management, surface-level analytics, and basically anything that an AI agent can do.
F Prime investor Abdul Abdirahman added that general purpose vertical software “without its own data moat” is no longer popular, and Igor Ryabenky, founder and managing partner of AltaIR Capital, elaborated on this point. He said investors aren't interested in products that lack depth.
“If the differentiation is primarily in the UI, [user interface] “That's not enough anymore. Barriers to entry have been lowered, making it much harder to build a real moat,” he said.
He said new entrants to the market now need to build around “true ownership of their workflow and a clear understanding of the problem from day one.” “Large codebases are no longer an advantage. What's more important is speed, focus, and the ability to adapt quickly. Pricing also needs to be flexible. A rigid per-seat model would be difficult to defend, but a pay-as-you-go model makes more sense in this environment.”
tech crunch event
San Francisco, CA | October 13-15, 2026
Jake Saper, general partner at Emergence Capital, also had thoughts on ownership. To him, the difference between cursors and Claude code is the “canary in the coal mine.”
“One person owns the developer workflow and the other person just executes the tasks,” Saper continued. “Developers are increasingly choosing execution over process.”
He said products that have “workflow stickiness,” meaning they try to attract as many human customers as possible to continue using the product, can struggle as agents take over the workflow.
“Before Claude, it was a strong moat to have humans do work within software, but who cares about human workflow if an agent is doing the work?” he told TechCrunch.
We also believe that integration is becoming less popular, especially as Anthropic's Model Context Protocol (MCP) makes it easier than ever to connect AI models to external data and systems. This means no one has to download multiple integrations or build their own customer integrations. Just use MCP.
“Being a connector used to be a moat,” Saper said. “It will be put into practice soon.”
Abdirahman also said, “Over time, workflow automation and task management tools that allow human work to be coordinated are no longer fashionable when agents simply perform tasks,” Abdirahman said, citing the example of primarily publicly traded SaaS companies whose stock prices have fallen as new AI-native startups with better and more efficient technology emerge.
Ryabenky said SaaS companies currently struggling to raise capital are easy to imitate.
“General-purpose productivity tools, project management software, basic CRM clones, and thin AI wrappers built on top of existing APIs fall into this category,” he said. “If a large part of the product is an interface layer without deep integration, proprietary data, or embedded process knowledge, a strong AI-native team can rebuild it quickly. That's what makes investors cautious.”
At the end of the day, what remains appealing about SaaS is the depth and expertise with which tools are built into critical workflows. He said companies should now consider deeply integrating AI into their products and update their marketing to reflect that.
“Investors are reallocating capital to companies that own workflow, data and domain expertise,” Ryavenky said. “And we move away from products that can be copied effortlessly.”