What VCs and investors are not looking for in SaaS?
Today, I read a TechCrunch article about what investors are no longer looking for in SaaS, or rather, what to avoid if you don't want to lose their interest.
The red flags were:
Too easy to replicate – light AI wrappers, generic horizontal tools, basic CRM clones, generic productivity or project management tools.
No real depth – products where differentiation is mostly UI and automation, anything without proprietary data, surface-level analytics.
Becoming obsolete – workflow automation tools that coordinate human work (agents are taking over), integrations as a moat (MCP is making connectors a commodity), and "workflow stickiness" products trying to keep humans inside their software.
The core message was that if an AI-native team can rebuild your product quickly, investors won't bet on it.
So, given all of that, how are you differentiating your SaaS product in the age of AI?
Because let's be honest: almost everything can be replicated now.
Replies
Almost everything can be replicated now, but the winners SaaS build real workflow ownership + proprietary data/agents + domain expertise from day one...
Are you going vertical/deep (e.g., industry-specific), building data moats, or shifting to agentic/infra plays? What's your biggest worry about replication right now?
minimalist phone: creating folders
@cathcorm To be honest, I do not focus on building SaaS as I am primarily a marketer, but I think that more specific solutions are about to win a market. E.g. my friend is programming some solutions for dentists and the military, and the use of AI can be useful there, but not completely helpful, as some sort of information and data cannot be shared completely with LLMs due to policies that companies have set for them internally. But it is a custom-made solution, not a SaaS. :)
@busmark_w_nika gotcha! Honestly, it's less about someone cloning the UI/features (that's table stakes now) and more about how fast someone copies the data flywheel... if we don't keep accelerating our proprietary dataset + agent fine-tuning loop, a well-funded generalist could catch up by brute-forcing similar outcomes with bigger models.
We're leaning hard into industry-specific (procurement/RFP world right now), with custom agents trained on curated, non-shared data + deep workflow integration so users feel like they "own" the process instead of renting a tool.
I think the real shift isn’t “AI can replicate everything” — it’s that surface-level SaaS is dying.
If your product is just UI + API orchestration, yes, it’s fragile.
But defensibility today comes from:
1) Proprietary data loops
2) Embedded distribution inside a niche
3) Workflow ownership, not just automation
4) Speed of iteration with user feedback
AI lowers the cost of building.
It doesn’t lower the cost of understanding a specific user deeply.
In my experience, the moat is no longer the feature — it’s the insight layer and execution velocity.
Curious how others are building data or behavioral moats into their products.
minimalist phone: creating folders
@andylee0 But it seems that this can win only big companies that already have big budgets to manage this :)
@busmark_w_nika
I’d actually argue smaller teams might have an edge here.
Big budgets help with scale, but they often slow down iteration.
Tight feedback loops, niche focus, and fast shipping don’t require massive capital — just proximity to users.
AI makes building cheaper.
It doesn’t automatically create insight.
@busmark_w_nika @andylee0 "fast shipping don’t require massive capital" - It's really!
@busmark_w_nika I think the bar has definitely shifted from “building software” to owning context and data.
A lot of SaaS products used to win by improving workflows or UI, but AI makes replication much faster now. What seems harder to replicate isn’t the interface — it’s domain understanding and proprietary operational data collected over time.
In vertical markets especially, differentiation comes from solving very specific problems deeply rather than building horizontal tools. If a product becomes part of how decisions are made (not just where tasks are managed), it’s much harder to replace.
AI can rebuild features quickly, but rebuilding accumulated insights, integrations with real operator workflows, and trust within a niche still takes time.
minimalist phone: creating folders
@fernandocampos From all the statements, I am mostly aligned with this one. It feels like it is hard to compete with others in the industry who can replicate it easily, and that's why many bet on better marketing.
Super interesting, thanks for sharing, Nika!
I'm curious if "easy to replicate" could also be related to market validation?
For example, PostGod is essentially "the agency LinkedIn ghostwriting process for $8.99 instead of $1,200." Could someone copy it? Absolutely.
But if the market is proven (agencies making millions doing this), and we're just making it accessible... does replicability matter less?
So, in this case, is proven market + better economics enough, or do VCs only want net-new categories?
I might also just be thinking out loud and taking this too far 😆
minimalist phone: creating folders
@ruxandra_mazilu I am just thinking whether we are now tapping into distribution (marketing). Because it is a differentiator of the product (but in any other way). But the product itself is too hard to differentiate, esp. that ones that can be built with AI and replicate features of others. 🤔