Product

Maybe not?

Maybe not?

How to un-AI a product

Mariia Krutsko

Founding Designer

Apr 22, 2025

I recently came across a podcast from NNG called “Designing AI Experiences,” featuring Caleb Sponheim. At one point, they asked him what he’d put on a billboard if every designer saw it driving to work in the morning.

His answer? “Maybe not?”

The man is clearly done with chatbots popping up in every app.

“AI is not a silver bullet to solving any problem.” - Caleb says.

And he’s absolutely right. There are better ways to meet your users’ needs. At least maybe not with yet another chatbot.

That’s exactly what we realized at Querio.

We started as an AI chatbot for data. It was cool at first—fast responses, natural language, all that stuff. But over time, we realized: this is just one piece of the puzzle. People working with data need way more than a chat bubble to make real decisions.

And let’s be clear—we’re not anti-AI. When used thoughtfully, it can be chef’s kiss: faster workflows, amazing insights, saving people’s time. But here’s the thing:

Too many products are adding AI just because they can—not because they should.

When AI Took Over Querio 😬

Like a lot of teams, we got a little too excited. We went all in on AI: chatbot, instant answers, AI chart builder, AI summary, AI everything—you name it. It sounded impressive. It even looked impressive.

But then… reality hit.

Users were confused. Some were low-key overwhelmed. Others didn’t trust the results because they couldn’t see where the info was coming from.

We had to face it: AI wasn’t making things better. It was just making things noisier.

So we…🥁… started to un-AI our product.

What Does It Mean to Un-AI a Product?

It means taking a breath and asking:

We realized AI should support the experience—not be the main character.

So we made some changes:

  • We put aside the chatbot and focused on showing and editing the output.

  • We started showing more of the logic behind the scenes, with editable steps.

  • We added a Knowledge Base where users define their own context and data rules.

  • We added Notebook with SQL and Python where users could write it all by hand or have Querio do the lift

Bottom line? We put the power back in our users' hands! But still keeping AI as an option.

The Un-AI Checklist

Next time when designing an AI feature ask yourself:

  1. Is AI solving a real user problem?

    Or are we just adding it because it sounds cool?

  2. Can users understand and control what the AI is doing?

    No mystery boxes. Give people visibility and control.

  3. Is the AI making things faster, or just fancier?

    A feature that looks cool but slows you down? Immediate red flag 🚩.

  4. Is this feature helping users do what they want to do?

    Let users lead, AI follows.

  5. Are users getting trust issues from AI?

    Transparency always wins.

Finally

We’re not saying AI is bad. We’re saying maybe not—not for every interaction, not in every corner of your product.

Querio

Query, report and explore data at technical level.

2025 Querio Ltd. All rights reserved.

Solutions

Our partners

Querio

Query, report and explore data at technical level.

Solutions

Our partners

2024 Querio Ltd. All rights reserved.

Querio

Query, report and explore data at technical level.

2025 Querio Ltd. All rights reserved.

Solutions

Our partners