Yannis Antonoglou

So these are systems that can use tools, they can code, they can use MCPs, which is an obstruction over software, so they can actually take actions and complete tasks.

excuse me, complete tasks end-to-end.

This is kind of like what we're building and this is why reinforcement learning and our backgrounds, like both my background and my co-founder's background is like so kind of like well aligned with like the systems that we're building.

Now, the question is like, how can we compete to the big labs?

And I think that, you know, we've,

Building models require a lot of resources and a lot of talent.

Reflection is one of the few places that, given the round that we raised, we've actually managed to get a lot of resources.

And we also attracted some of the best talent in the world.

So actually, people who've done their thinking models, both in DeepMind and OpenAI, led the efforts there.

They've actually joined Reflection.

We have people who...

We're like one of the leads in pre-training at Gemini who kind of lead our pre-training here.

So we have people who build the most impactful models and systems at the big labs who've actually come together with us to just ensure that this technology becomes available and accessible to people.

They really believe and they really share our dream and vision of open weight models.

At the same time, you know, like we are...

You can see that there are many frontier open models coming out of China.

And the Chinese labs, they don't have the same resources that maybe the closed labs have.

But at the same time, that hasn't stopped them from building some really, really powerful models like DeepSea Core Kimi.

So we are really bullish that we'll just repeat the same in the West.

Yes.