Investors spill what they aren’t looking for anymore in AI SaaS companies
Source: TechCrunch
Overview
Investors have poured billions into AI companies over the past few years, but not all AI startups attract attention. While many firms now brand themselves with “AI,” certain ideas have fallen out of favor. TechCrunch spoke with VCs to learn what investors aren’t looking for in AI software‑as‑a‑service (SaaS) startups anymore.
What Investors Are Looking For
- AI‑native infrastructure – platforms built from the ground up for AI workloads.
- Vertical SaaS with proprietary data – solutions that leverage unique data sets to create defensible moats.
- Systems of action – tools that help users complete tasks rather than just provide information.
- Deeply embedded mission‑critical workflows – products that become indispensable within a specific domain.
Aaron Holiday, managing partner at 645 Ventures, highlighted these categories as the current sweet spots for investors.
What Investors Aren’t Looking For
Boring or Low‑Moat Categories
- Thin workflow layers
- Generic horizontal tools
- Light product‑management solutions
- Surface‑level analytics (anything an AI agent can already perform)
Lack of Proprietary Data or Depth
- Generic vertical software without proprietary data moats (Abdul Abdirahman, F‑Prime)
- Products whose differentiation lives mainly in UI and automation (Igor Ryabenky, AltaIR Capital)
“If your differentiation lives mostly in UI and automation, that’s no longer enough. The barrier to entry has dropped, making a real moat much harder.”
Replicable SaaS Products
- Generic productivity tools
- Project‑management software
- Basic CRM clones
- Thin AI wrappers built on top of existing APIs
Ryabenky warned that these can be quickly rebuilt by AI‑native teams, making investors cautious.
Key Insights from Investors
Workflow Ownership
Jake Saper, general partner at Emergence Capital, contrasted Cursor and Claude Code as a “canary in the coal mine.”
- Cursor owns the developer’s workflow.
- Claude Code merely executes tasks.
“Developers are increasingly choosing the execution over process.”
Saper noted that products relying on “workflow stickiness” (continuous human use) face an uphill battle as AI agents take over tasks.
Integration Moats Are Fading
Anthropic’s Model Context Protocol (MCP) simplifies connecting AI models to external data and systems, reducing the need for custom integrations.
“Being the connector used to be a moat. Soon, it’ll be a utility.”
Pricing Models
- Rigid per‑seat pricing is harder to defend.
- Consumption‑based models make more sense in the current environment.
Speed and Focus Over Massive Codebases
“Massive codebases are no longer an advantage. What matters more is speed, focus, and the ability to adapt quickly.”
Implications for SaaS Startups
- Build real workflow ownership – design products that deeply understand and solve a specific problem from day one.
- Embed proprietary data – develop data moats that are hard to replicate.
- Integrate AI deeply – go beyond surface‑level features and embed AI into core functionality.
- Adopt flexible pricing – consider consumption‑based models rather than fixed per‑seat fees.
- Prioritize speed and adaptability – focus on rapid iteration and the ability to pivot as the market evolves.
Conclusion
Investors are reallocating capital toward SaaS businesses that own workflows, data, and domain expertise, while moving away from products that can be copied with little effort. Startups that embed AI deeply, leverage proprietary data, and demonstrate clear, mission‑critical value are more likely to attract funding in the evolving AI SaaS landscape.