With Everyone Talking AI, Don’t Miss Actual User Needs
“We need to add AI.” “Can we have some AI automation in this process?” “Can we build a chatbot?” “What’s our Gen AI roadmap?”
If you're a product manager, owner, or any member in 2025, chances are you've heard some version of this in your meetings. With AI taking center stage — from GPTs writing code to copilots assisting every function — the pressure to integrate artificial intelligence into products is real, loud, and often misdirected.
But amidst this wave, one question keeps me grounded: Does this solve a real user problem?
The temptation of trend-driven roadmaps
AI is exciting. It's powerful. And yes, in many cases, it's transformative. But here’s the catch: not every product needs AI — and certainly not all at once.
Too often, I've seen teams race to embed AI features: - A predictive dashboard that users never asked for - A recommendation engine that creates more confusion than clarity - A chatbot that replaces help articles no one wanted to read anyway
Why? Because everyone else was doing it!
Re-centering around the user
A few guiding principles have helped me resist this pressure:
1. Don’t start with tech –– start with friction. What are users struggling with? Where do they spend time inefficiently? Start there. AI might be the answer — but sometimes a better form, simpler flow, or a tooltip does the job.
2. Validate desire, not just feasibility. Just because your engineers can build it, doesn’t mean your users will want it. Use quick-and-dirty user testing. If a prototype doesn’t spark curiosity or clarity, it’s a sign.
3. Look for patterns, not noise. AI is great at scale — which means it’s best applied to repetitive patterns. Use data to find where automation, prediction, or summarization makes sense.
Listening over launching
Recently, during a customer interview, a user told me, “I don’t care if it’s AI or not — I just want it to save me time.” That stuck with me. Most users won’t say “Give me AI.” They’ll say, “I’m tired of doing this manually,” or “I wish this knew what I needed before I had to search.” It’s our job to translate that into product design — not to impose buzzwords.
What can actually work: use cases
AI to auto-classify support tickets — only after users shared how overwhelming categorization was.
Summarization of reports — after noticing users copy-pasting data into ChatGPT themselves.
Saying no to an AI assistant idea — because user interviews showed they preferred visual dashboards over chats.
Each case was driven by a real, validated user need — not just a Gartner trendline.
Build less AI — solve more roblems
AI should be an enabler, not a feature checklist. At the end of the day, your users don’t care about the underlying magic. They care that your product fits into their life just a little more seamlessly than it did yesterday.
So, the next time someone says, “We should add AI…,” try asking: “Which user problem are we solving — and is AI the best way to solve it?”