Despite the continued flood of venture capital into artificial intelligence — with billions allocated annually to AI-driven companies — not every category within the sector commands investor enthusiasm. A growing number of software-as-a-service concepts, once considered viable bets, are now viewed as uninvestable by leading venture firms. The distinction between what attracts capital and what repels it has never been sharper.
To understand where investor appetite has gone cold, TechCrunch canvassed several prominent venture capitalists for their candid assessments. Their responses reveal a consistent theme: depth, proprietary data, and genuine workflow ownership are now the non-negotiable criteria — and most early-stage SaaS concepts fail to clear that bar.
According to Aaron Holiday, managing partner at 645 Ventures, the categories that still generate investor excitement include AI-native infrastructure, vertical SaaS backed by proprietary data, platforms embedded in mission-critical workflows, and so-called systems of action — tools that actively help users complete tasks rather than merely organizing them. What no longer qualifies, in his assessment, is a long and familiar list.
"Startups building thin workflow layers, generic horizontal tools, light product management, and surface-level analytics — basically, anything an AI agent can now do."
That last phrase is the operative concern. As AI agents grow more capable, they increasingly displace the lightweight software layers that once justified their own product categories. Abdul Abdirahman, an investor at F Prime, noted that generic vertical software "without proprietary data moats" has lost its appeal, a sentiment echoed and expanded upon by Igor Ryabenky, founder and managing partner at AltaIR Capital.
Ryabenky's critique centers on product depth — or the absence of it. In his view, a polished interface and a degree of automation are no longer sufficient to establish a defensible market position.
"If your differentiation lives mostly in UI [user interface] and automation, that's no longer enough. The barrier to entry has dropped, which makes building a real moat much harder."
The implication for founders is significant. Ryabenky argues that new market entrants must orient their products around genuine workflow ownership and a deep understanding of customer problems from the very first iteration. He also challenged long-held assumptions about the advantages of scale.
"Massive codebases are no longer an advantage. What matters more is speed, focus, and the ability to adapt quickly. Pricing also needs to be flexible: rigid per-seat models will be harder to defend, while consumption-based models make more sense in this environment."
The question of workflow ownership emerged as a central fault line in investor thinking. Jake Saper, general partner at Emergence Capital, pointed to the divergence between Cursor and Claude Code as an instructive case study in how the market is bifurcating.
"One owns the developer's workflow, the other just executes the task. Developers are increasingly choosing the execution over process."
Saper argues that products built around "workflow stickiness" — the strategy of embedding as many human users as possible into a product's daily routines — face a structural threat as AI agents begin performing those same tasks autonomously. The moat that once made such products compelling is eroding rapidly.
"Pre-Claude, getting humans to do their jobs inside your software was a powerful moat, but if agents are doing the work, who cares about human workflow?"
Saper also raised concerns about integration-layer businesses, which have historically derived value from connecting disparate tools and data sources. With Anthropic's model context protocol (MCP) making it significantly easier to connect AI models to external systems, the competitive advantage of being a connector has diminished considerably.
"Being the connector used to be a moat. Soon, it'll be a utility."
Abdirahman reinforced this point by flagging the declining relevance of workflow automation and task management tools — those designed to coordinate human work. As AI agents become capable of executing tasks outright, the coordination layer loses its purpose. He cited public SaaS companies whose stock valuations have declined as AI-native competitors emerge with more efficient, purpose-built alternatives.
Ryabenky offered a practical framework for identifying the types of SaaS businesses that are struggling most to secure funding today. The common thread, he said, is replicability.
"Generic productivity tools, project management software, basic CRM clones, and thin AI wrappers built on top of existing APIs fall into this category. If the product is mostly an interface layer without deep integration, proprietary data, or embedded process knowledge, strong AI-native teams can rebuild it quickly. That is what makes investors cautious."
For existing SaaS companies navigating this environment, Ryabenky's prescription is direct: integrate AI deeply into the product and ensure that the company's positioning reflects that shift. Surface-level AI adoption will not satisfy investor scrutiny.
The broader reallocation of capital is already underway. As Ryabenky summarized, the firms attracting investment are those that have built durable advantages through data ownership, domain expertise, and deep process integration — while those offering little more than a repackaged interface are being passed over with increasing regularity.
"Investors are reallocating capital toward businesses that own workflows, data, and domain expertise. And away from products that can be copied without much effort."
