Vector

Felix Tay, Chief AI Officer of Unity, you launched Vector a year and a half ago.

What did you believe about the product at the announcement that you no longer believe now?

Well, let's just say before I touch upon the product, probably the more important bit is when you're launching something new in which you believe is transformative, but you don't have any empirical evidence outside of your own lap of what that thing says.

So if there's one thing that I guess I have now that I didn't have back then is the empirical evidence about the fact that the customers actually love it.

Well, yeah, well, an upside surprise is a very positive thing to be wrong about.

And so I've been watching Vector's progress as a customer, both on the demand and the supply side.

We've been active users.

What was the hardest thing to get working that you didn't expect? Because what you're describing is totally true.

So, uh, thanks to Rick Faulk and the team at Locus Robotics for hosting me.

Uh, they've got, I think, eight zillion-plus picks that they've assisted with, and it was great to see, uh, all of their, um, Locus Origin and, and Vector robots being prepared for the holiday rush.

So they had hundreds of robots that they're shipping out practically daily now, uh, to help their customers flex capacity, and, uh, that was really interesting.

And, you know, I got to see the, the new gripper that's going onto, uh, Locus Array, their mobile manipulator.

Keyword is deterministic and predictable.

Vector is semantic and unpredictable.

And that's before we get to the machinery that makes either one stop misfiring, which is where secondary keys come in.

Selective matching.

So you might have an index per customer, an index per team, an index per chat history, an index per agent, and that means that your system has to be able to store a very, very large volume of unique indices and be able to search over them with a high degree of accuracy.

And if you're to store all of those in one index, one, you have to pay the cost of keeping that live at all times, and then two, you have to deal with the accuracy degradation in vector search when all, everything is in one index.

Vector search doesn't do well with being partitioned, and so the best thing to do is hard partition the data, keep it actually in logically separate indexes.

You can't filter the search at query time.

You have to do it at index time.

And so Context One is just trying to fit into that paradigm.

So I guess looking again what actually makes up this sub-agent, I think there are, like, four tools.

You got Vector+BM25, Hybrid Search, Regex/Grep, Redocument, and Prune Chunks, and there could be a lot to cover there, but I think more to the point of, like, why those four specifically?