A customer asked what makes us different from a chatbot. We added eyes

Product designPersonalityTrade show
Updated 2026-09-13

Putting two small displays on a voice toy as eyes was not about looking nice. It was about getting a stranger to walk over.

The sentence that changed my mind

I had not planned to build eyes. Anyone doing hardware has the same reflex. One more display means one more driver and more cost, so cut it and put the money into something functional.

Then a customer asked me a question. If it has no eyes, what makes it different from a chatbot on a phone?

I did not have an answer for a few seconds.

Of course we differ in plenty of ways. Different knowledge base, different scenario, different purpose. But to a stranger who knows none of that, sitting silently on a table, what is supposed to make them want to come closer and ask what it is?

Function makes a product useful. A face makes it a character. Those are separate jobs.

At a trade show, attention beats a pitch

I have watched this play out many times. A row of AI speakers on a stand, people scan across and keep walking. Put something there that moves, that looks back at you, whose expression changes, and a lot of people slow down and ask what it is.

That approach is worth more than any pitch, because they initiated it.

Voice products have a built-in problem. When they are not talking, they look like a speaker. Eyes address the silent state, so that before it has said anything it already reads as something with a bit of life in it.

What the expression system actually does

Having decided to build eyes, I did not want a few looping frames of animation. Loops read as dead within two seconds and the thing becomes a machine again.

The expressions are bound to state. The eyes move with the rhythm of speech while talking, there is a distinct look while thinking, and different moods render differently. All designed, none random.

I also added a set of offline interactions, and that is the part I care about most. Stroke it and a good-luck sequence triggers, with traditional zodiac patterns appearing frame by frame. Shake it and the screen shows a dizzy face with unfocused eyes, exaggerated on purpose. None of this needs a network, so it works at any moment.

Why spend time here? Because personality accumulates out of details. A user may forget what the device looked up for them. They remember that they shook it and it pulled a face. That face is what makes it feel like something with a temper rather than a cold appliance.

What I took from it

I have made a lot of utility-first calls and cut plenty of designs that were attractive and useless. Eyes are one of the few times I deliberately put emotional value first.

Every time we take the device out, the eyes are the first thing anyone notices. Before any introduction, someone comes over. That result says the customer was right.

Function decides whether people can use it. Character and appearance decide whether they want to pick it up. The second is sometimes a precondition for the first.

Want to try the hardware behind these notes?

Six AI hardware platforms are in mass production. Tell us your market and channel and we send an evaluation sample.

Request a sample