“We’re adding AI to our product.”

I hear this weekly from founders. My first question: Why?

Usually the answer is some version of “because everyone else is” or “investors expect it.”

That’s the mistake.

The AI Feature Nobody Wanted

A founder I know spent $45,000 adding an AI assistant to their project management tool. Three months of development. Hired a contractor. The works.

Usage after launch: 3% of users tried it. 0.4% used it twice.

The feature added nothing to retention, nothing to conversion, nothing to revenue. It just existed.

Why? Because customers never asked for it. They wanted better mobile apps and faster syncing. But AI sounded cooler.

The Right Way to Think About AI

AI is a technology. Not a strategy.

Your strategy is solving customer problems profitably. AI is sometimes the best tool for that. Often it’s not.

Before any AI project, answer these:

1. What specific customer problem does this solve?
2. Have customers asked for this solution?
3. Why is AI better than non-AI approaches?
4. What’s the expected revenue or retention impact?

If you can’t answer all four, you’re doing AI for vanity.

Where AI Actually Helps Startups

Internal operations. Not customer features.

AI shines at:

• Summarizing documents and emails
• Drafting first versions of content
• Analyzing data and spotting patterns
• Answering repetitive questions
• Automating tedious workflows

These save real money. Real time. Real headaches.

Customer-facing AI is harder. Users have high expectations. They’ve used ChatGPT. Your chatbot will feel dumb by comparison.

The Honest Assessment

For most startups under $5M revenue, AI should be:

High priority: Internal productivity tools, content creation, data analysis

Medium priority: Customer support automation, personalization

Low priority: New AI product features, chatbots, “AI-powered” anything

Build your core product first. Add AI where it creates measurable value. Not where it sounds impressive on a pitch deck.

The Exception

If AI is your product, different story. But even then, the question is customer value, not technology coolness.

Nobody cares that you used GPT-4 Turbo with function calling and RAG. They care if your product solves their problem better than alternatives.

My Rule

I won’t start an AI project without a customer quote explaining why they need it.

Not “customers would love this” from the product team. An actual quote from an actual paying customer saying “I wish your product could X.”

No quote, no project. Simple.

Keeps us focused on value instead of hype.