Back to all essays
Own Your TechGeorge's Takes

Why We Started With Character

·1 min read
George Pu
George PuBuilds in AI

28 · Toronto · Building to own for 30+ years

Why We Started With Character

The current state of frontier AI is owned by a handful of labs.

Led by two American companies with closed models you can't see inside.

And a wave of Chinese labs shipping open weight models that are, honestly, very good and capable.

I've used all of them. The capability is real.

But something is missing.

Most of the open weight models are robotic to talk to, doesn't matter what the size is.

Smart, fast, capable - and cold.

They answer like a manual.

They'll make something up before they admit that they don't know.

There's no one home.

For a while the choice looked like this: capable but closed, or capable but cold.

Pick one.

I never believed that was the real choice.

The thing that's actually scarce isn't capability.

Capability is getting cheaper every month - another lab, another open release, another benchmark beaten.

What's scarce is a model you actually want to talk to.

One with a point of view.

One that tells you the truth, including when the truth is 'I don't know.'

That's the bet we've made from the start.

Not 'build the smartest model.'

We can't outspend the labs, and we're not going to try.

Build the one with character.

The one that's honest. The one you'd choose to keep around.

So that's where we started.

Not with a 400 billion parameter flagship.

With a small model - 4B - that you can run on your own laptop, that's free, and that has a personality we shaped on purpose.

It's small.

It won't out-reason the giants, and I'm not going to pretend it does.

But it's ours, it's open, and it's honest.

And it's the first proof of the thing we've believed all along -

That character is the moat, and you can build it in on purpose.

This is Vinci 1.0. The smallest version of the idea.

The rest of the family is coming.

Try it on getsimpledirect.com on June 30th, 2026.

Share this: