Traders have been pouring billions into AI firms over the previous few years, because the know-how continues to carry sway within the Valley and thus the world. However not all AI firms are grabbing investor consideration.
Certainly, even because it appears each firm as of late is rebranding to incorporate “AI” in its identify, some startup concepts are simply not in favor with traders. TechCrunch spoke with VCs to be taught what traders aren’t searching for in AI software-as-a-service startups anymore.
Well-liked SaaS classes for traders now embody startups constructing AI-native infrastructure, vertical SaaS with proprietary information, programs of motion (these serving to customers full duties), and platforms deeply embedded in mission-critical workflows, in keeping with Aaron Vacation, a managing accomplice at 645 Ventures.
However he additionally gave an inventory of firms which are thought of fairly boring to traders as of late: startups constructing skinny workflow layers, generic horizontal instruments, mild product administration, and surface-level analytics — principally, something an AI agent can now do.
Abdul Abdirahman, an investor on the agency F-Prime, added that generic vertical software program “with out proprietary information moats” is not widespread, and Igor Ryabenkiy, a founder and managing accomplice at AltaIR Capital, went deeper on that time. He stated traders aren’t occupied with something, actually, that doesn’t have a lot product depth.
“In case your differentiation lives principally in UI [user interface] and automation, that’s not sufficient,” he stated. “The barrier to entry has dropped, which makes constructing an actual moat a lot more durable.”
New firms coming into the market now must construct round “actual workflow possession and a transparent understanding of the issue from day one,” he stated. “Large codebases are not a bonus. What issues extra is velocity, focus, and the power to adapt shortly. Pricing additionally must be versatile: inflexible per-seat fashions can be more durable to defend, whereas consumption-based fashions make extra sense on this atmosphere.”
Techcrunch occasion
San Francisco, CA
|
October 13-15, 2026
Jake Saper, a basic accomplice at Emergence Capital, additionally had ideas on possession. To him, the variations between Cursor and Claude Code are the “canary within the coal mine.”
“One owns the developer’s workflow, the opposite simply executes the duty,” Saper continued. “Builders are more and more selecting the execution over course of.”
He stated any product coping with “workflow stickiness” — which means attempting to draw as many human prospects as attainable to repeatedly use the product — would possibly discover themselves in an uphill battle as brokers take over the workflow.
“Pre-Claude, getting people to do their jobs inside your software program was a robust moat, but when brokers are doing the work, who cares about human workflow?” he instructed TechCrunch.
He additionally thinks integrations have gotten much less widespread, particularly as Anthropic’s mannequin context protocol (MCP) makes it simpler than ever to attach AI fashions to exterior information and programs. This implies somebody doesn’t must obtain a number of integrations or construct their very own buyer integrations; they will simply use the MCP.
“Being the connector was once a moat,” Saper stated. “Quickly, it’ll be a utility.”
Additionally, the “workflow automation and activity administration instruments that allow the coordination of human work develop into much less vital if, over time, brokers simply execute the duties,” Abdirahman stated, citing examples, primarily public SaaS firms whose shares are down as new AI-native startups come up with higher, extra environment friendly know-how.
Ryabenkiy stated the SaaS firms struggling to lift proper now are those that may simply be replicated.
“Generic productiveness instruments, challenge administration software program, fundamental CRM clones, and skinny AI wrappers constructed on high of present APIs fall into this class,” he stated. “If the product is generally an interface layer with out deep integration, proprietary information, or embedded course of data, sturdy AI-native groups can rebuild it shortly. That’s what makes traders cautious.”
General, what stays enticing about SaaS is depth and experience, with instruments embedded in important workflows. He stated firms ought to proper now look into integrating AI deeply into their merchandise and replace their advertising to replicate that, Ryabenkiy continued.
“Traders are reallocating capital towards companies that personal workflows, information, and area experience,” Ryabenkiy stated. “And away from merchandise that may be copied with out a lot effort.”
Keep forward of the curve with NextBusiness 24. Discover extra tales, subscribe to our publication, and be part of our rising group at nextbusiness24.com
