Look at any enterprise SaaS category in 2026 and you will see the same AI features in every product. Every CRM has an AI email writer. Every project management tool has an AI task summarizer. Every analytics platform has an AI insight generator.
The features are interchangeable. The marketing copy is interchangeable. Even the UX is interchangeable—it is always a sparkle icon that opens a side panel with a chat interface.
When everyone in the field is facing the same direction, the most valuable position is the one nobody else is looking at.
The Herd Instinct in AI Product Development
Herd behavior in SaaS is not new. Every few years, a new technology creates a wave that every product in every category rides simultaneously. Mobile-first. Cloud-native. Real-time collaboration. And now, AI-powered everything.
The pattern is always the same:
- Pioneer phase. One or two companies build genuinely novel AI features that create real value.
- Fast-follow phase. Every competitor copies the surface-level implementation within 6 months.
- Feature parity phase. AI features become table stakes. No product can differentiate on them.
- Disillusionment phase. Customers realize that every product's AI features are equally mediocre.
- Divergence phase. The companies that invested in unique, deep AI capabilities pull ahead.
We are currently in the transition from phase 3 to phase 4. The disillusionment is setting in. Customers are asking: "Every tool has AI now, so why am I still doing the same amount of work?"
What the Herd Is Building
If you survey the top 50 enterprise SaaS products, their AI features cluster into five categories:
- Content generation. Write this email, generate this report, draft this proposal.
- Summarization. Summarize this thread, recap this meeting, digest this document.
- Search enhancement. Natural language search across your data.
- Chatbot interface. Ask questions about your workspace in a chat sidebar.
- Simple automation. If X happens, trigger Y using AI to determine the condition.
These are not bad features. They are just obvious features. Every product team looked at the same LLM capabilities and arrived at the same conclusions. The result is a sea of sameness.
Finding Your Contrarian Bet
A contrarian bet is not just doing something different. It is doing something that most smart people think is wrong, but that you have evidence suggesting is right.
Here are the frameworks we use to identify contrarian AI bets:
1. Look Where the Herd Is Not Looking
The herd is focused on generation (making AI create new content). The contrarian bet is on elimination (using AI to remove work entirely).
Generation adds a step to the workflow: the user now has to review and edit AI-generated content. Elimination removes steps from the workflow: the user no longer has to do the task at all.
Example: instead of using AI to draft a weekly status report (generation), use AI to eliminate the need for weekly status reports entirely by automatically tracking and surfacing progress to stakeholders (elimination).
2. Invert the Interface
The herd builds AI features as interactive tools: the user asks, the AI answers. The contrarian bet inverts this: the AI asks, the user answers.
An AI that asks good questions is more valuable than an AI that gives good answers. Because good questions surface information that the user did not know they needed to share, which leads to better decisions.
Example: instead of a chatbot that answers questions about your sales pipeline, build an AI that proactively asks "You have 3 deals closing this week with no activity in 5 days—should I flag these for your team's attention?"
3. Bet on Boring
The herd builds flashy AI features that demo well. The contrarian bet is on boring AI features that compound.
The most valuable AI features are the ones that save 30 seconds, 50 times a day. Not the ones that save 10 minutes once a week. Because 30 seconds times 50 times 250 working days is 104 hours per year. That is 2.5 working weeks reclaimed.
Example: instead of an AI-powered "insights dashboard" that produces impressive-looking charts, build AI that automatically fixes data quality issues (standardizing company names, deduplicating contacts, filling in missing fields). Boring. Invisible. Enormously valuable.
4. Solve the Problem Before the Problem
The herd builds reactive AI: it helps after the user encounters a problem. The contrarian bet is on predictive AI: it prevents the problem from occurring in the first place.
Example: instead of using AI to help users recover from a failed integration sync, use AI to predict sync failures before they happen and automatically adjust the sync configuration.
The Sheep Test
Before investing engineering resources in an AI feature, we apply the Sheep Test:
If you described this feature to your top 5 competitors, would they say "oh, we are building that too"?
If yes, you are following the herd. The feature might be necessary for parity, but it will not differentiate your product.
If they say "why would you build that?"—congratulations, you might have a contrarian bet. Now validate it.
Validation Without Consensus
The hardest part of contrarian bets is that you cannot validate them by asking the market. By definition, the market disagrees with you. If you ask customers whether they want the feature, most will say no—because they cannot imagine what they have not seen.
Instead, validate contrarian bets with:
-
Behavior observation. Watch what users actually do, not what they say they want. If users spend 20 minutes every Monday morning manually categorizing new data entries, there is a contrarian bet in automatic categorization—even if no user has ever asked for it.
-
Small experiments. Build the simplest possible version and measure impact on a single metric. If automatic categorization reduces Monday morning data processing time by 80%, you have your validation.
-
Second-order effects. The value of contrarian bets is usually one step removed from the obvious. Automatic categorization does not just save time—it means data is categorized consistently, which means reports are more accurate, which means decisions are better. Measure the second-order effect.
The Long Game
Following the herd is safe in the short term. Your product has AI parity. Your marketing says "AI-powered." Your board is happy.
But in the long term, herd features commoditize. They become checkboxes. They stop being a reason to choose your product and start being a reason not to leave.
Contrarian bets are risky in the short term. They are hard to explain. They do not demo as well. Your board might not understand them.
But in the long term, contrarian bets become moats. They are hard to copy because competitors dismissed them. They compound because they solve real problems rather than performing AI theater.
The sheep on the hill sees the same grass as every other sheep. The one that wanders to the next valley finds the meadow no one else knows about.
Do not follow the herd. Find your meadow.
